Into the Gateway-verse: Bootstrapping a Multi-Gateway Ecosystem

Thank-you very much for this response! It does provide MUCH clarity, and enable there to be a conversation based on material information.

I think the primary point of miscommunication is around PNI’s Gateway-as-a-Service (GaaS) model and appstake sharing. The GaaS model was based on using PNI’s backend infrastructure with others building custom front-ends. I know that myself and others did not find that a compelling model for many reasons. I believe that is why so few folks showed interest in gateways when that was the only available model.

However, once it was announced that the GaaS model was NOT the future, but instead appstakes would be independently operated, there has been significantly more interest across the community from what I have seen. An appstakes model is a completely different beast as there isn’t a dependency on PNI (like the GaaS model would introduce).

It seems that interest for the GaaS model (dependent gateways) and the appstakes model (independent gateways) were being unintentionally conflated by PNF as the same thing. Because Nodies and BargSystems showed interest in GaaS, they were part of the internal conversations regarding appstakes, which I feel was an oversight and did provide them with preferential access to formulate their strategy and now access to uncontested funding.

I first heard about the appstakes model after ETHDenver and, and immediately expressed interest in the community call and we were told by both PNI and PNF that more information would be coming. There were little updates occasionally, but nothing that signaled there was conversations happening where interested parties could be involved. I have witnessed others expressing interest the independent gateways as well. Unfortunately there hasn’t been actionable information since, and I wrongly assumed POP’s were going to be used for the entire project, not just a small subset.

I also believe my concern about double founding the same roadmap is valid. Nodies received 6.9M POKT which has yet to be touched, and yet they have been operating with the same number of team members since February. Again, it confusing what the LeanPOKT proposal is funding if not their roadmap as indicated in their proposal and the conversation that followed.

I would also like reassurance that everything is crystal clear what they are building with any extra funding, so there isn’t a similar situation that happened with the LeanPOKT contribution. While PNI and the community was testing and improving the LeanPOKT open-souce client for everyone’s benefit, they launched their staking business entirely on a closed source client with a version of geo-meshing, unannounced to the community. While a savvy strategy, I personally saw that as the community contributing and later very lucratively rewarding one client, while another client was developed in that exact same time frame and used by their business to garner substantial market capture over their competitors.

Therefore, I feel it is in the best interest of everyone to ensure there is not the same blind spots in further funding. Everyone (including myself) have commended them for their LeanPOKT contribution and desired to reward their efforts, but it is unclear how much of the LeanPOKT funding went towards the open-source version and what went to the closed-source version they actually used for their business, since they were developed in the same time-frame covered by the LeanPOKT proposal. It would be a major loss if that same scenario played out with gateway contributions.

If they produce quality open-source tooling that has universal utility within the POKT ecosystem, then it is worth making it happen, though I would hope there be vigilant PMing with detailed technical specs that can be shared with the community, defined deliverables and roadmap, along with a clear SLA/contract with BaaS Pool LLC. I say this not at all to micro-manage, as PNF is extremely competent, I’m just being honest about blind spots we have had in the past and my perspective.

I do also believe that $300k is too much considering their LeanPOKT proposal, who’s value was partially predicated on their roadmap (which included the gateway), so I would suggest that it be taken into account, and provide more POP opportunities for others (which Nodies could apply for as well).

I leave this feedback for PNF’s consideration :+1:


Hi @shane, I believe @JackALaing already clarified a number of your questions, and we’re glad to see you’re aligned with the value proposition of gateways, as well as what an undertaking this will be. We also wanted to share our perspectives on a few things.

First, I would like to point out that the Quote you pulled was from a Pre-proposal, the details of which were quite different from the actual proposal, which the DAO actually voted for and approved. For one thing, you will find that the quote you pulled was not included in the passed proposal.

Second, there appears to be some confusion regarding the timelines involved. The development of LeanPokt by the Poktfund team, led by @poktblade, began in February 2022 and was merged into Pocket Core in December 2022. Subsequently in February 2023, we submitted a reimbursement request for the work completed by our team during that period. Currently, we are seeking funding for the upcoming 6 months of work. Therefore, if we consider the start dates of these two asks, there is actually a gap of over a year between them. To be clear, The LP Proposal (and its subsequent passage) was not intended to fund future projects. LeanPOKT was also never intended to be an all-inclusive grant for all the work we perform at Pocket. Generally speaking, we consistently do pro-bono work for Pocket that adds value to the ecosystem, but considering the scope of work required for the gateway, it made sense for us to seek funding.

Furthermore, there have been changes in personnel over time. Some of our team members who worked on LP have transitioned to our staking operation, which is unrelated to our gateway operations. Some, like @poktblade, have remained; while others, including myself, have joined as new team members specifically for Nodies DLB. I recently began my role as a business development representative for our gateway.

Thank you for transcribing from the Community Call. I believe this was misconstrued, so allow me to clarify. @poktblade’s intention was to highlight that in the event Pocket Network is unable to handle specific requests (for instance, if a customer needs websockets, but Pocket is not yet capable of handling those requests), NodiesDLB will still handle the requests to meet the client’s needs. This will allow us to incentivize DApps to use our platform (developer productivity is a common value prop), help provide market data to the protocol team and shape the roadmap, and ultimately pipe this to Pocket once it is supported. If the focus is demand, then we need to be competitive in our product offerings by strengthening our value propositions and providing DApps with what they need instead of turning them away. This in return will indirectly provide traffic to Pocket as well, as clients hardly want to use multiple providers for blockchain development. If you take a look at many other RPC providers, this is a common strategy in order to capture demand.

This is a mischaracterization based on incorrect assumptions. First, we released LP well before Nodies began running nodes, and we never stopped working on LP even as “PNI and the community” were testing the beta. Second, it is true that Nodies’ commercial staking operation began in September 2022 (and some small nodes for ourselves, friends, and family prior), and we were using a software akin to the Geo-Mesh. There were conversations to open source this work as well, but it was in alpha testing during a time when we had a relatively small amount of nodes and limited data to test. Even then, it would suffer from outages fairly often, bringing down some of our nodes with it. It took us a while to get the software to a stable enough state. We prioritized the network’s health and security similar to how LeanPocket was originally closed source as advised by the protocol team. By that point, the PoktScan had already released their Geo-Mesh in October 2022 with a much more reliable and supported codebase, hence we didn’t see the need to release ours. The timeframe between all of this is really quite short and hardly gave us much market advantage. In fact, we already alluded to this in response to a comment you made back during LeanPOKT’s Preproposal -

Lastly, we have since switched over all of our nodes to Geo-mesh in the spirit of keeping consistent with the rest of the network and encouraging open sourceness within the ecosystem. In fact, we are the second and only other contributor to Geomesh. We’ve worked with them to identify bugs, optimizations, code reviews, and have been credited multiple times directly from the main developer @Jorge_POKTscan (PR 1, PR 2, and countless of private conversations on how to architect and code).

I hope this clears some things for you. We are excited to continue contributing towards enhancing and supporting Pocket Network in various aspects!


Hey @giantfrog, I don’t believe we have met before. I appreciate the response.

Oh, I’m aligned. So much so, I wrote the original ADR that the v1 integration is based on :stuck_out_tongue_winking_eye:

Pre-proposals are where the conversations happen for proposals… that is their intended design. You will find that with all the pre-proposal/proposal pairings, the real conversation for the proposal is in the first version (pre-proposal) and not the final version. It is a new perspective to try and divorce their relevancy to each other.

The final proposal for LeanPOKT was fast tracked and put to a vote after only 5 days, instead of the standard minimum of 7 days, precisely because the pre-proposal was the context, who’s conversation expanded over many weeks. Therefore I’d prefer to use the whole community conversation for context.

Unfortunately, that was not clear at the time because the proposal’s value was not predicated on measurable metrics, but more abstract concepts of value. Concepts that were partly built on notions that this proposal is worth it’s ask because of the value it would produce in the ecosystem (which thus far it has been unused). This is why I have always advocated for measurable metrics with every proposal, including LeanPOKT, so there would be universal clarity :sweat_smile:

It may be clear to you, but for myself, as a voter, outsider, and participant in the LeanPOKT proposal conversation, it is not clear. That is my point and why I provided past context.

As much I would like to take credit for transcribing, I must give it to Youtube :joy:

Yeah, I agree that if Nodies is able to capture customers for non-protocol features, it can indirectly lead to more traffic to POKT as well. I get the strategy.

I totally get that. Nodies began running nodes in August/September with geo-meshing, which means the development of the private client began before that, and active development of LeanPOKT was still underway between PNI and @poktblade in July/August, so it all overlaps. That is my point… the timing is very messy and confusing, which I why I believe it is relevant to today’s conversation. I also believe this kind of scenario should be avoided moving forward.

I would really appreciate it if we call a spade a spade and not try to now frame Nodies market capture as being an act of benevolence for the health and security of the network. From my understanding the geo-mesh code was never given to PNI to vet (LeanPOKT on the other hand was shared), nor did PNI ask to keep the geo-mesh concept private (they always encourage public discussion regarding LeanPOKT), so these are not equivalent scenarios. It was a great business move, and it took POKTScan months to reverse engineer in an open-source fashion, so I’d prefer we leave it at that.

While I can respect that the narrative at Nodies is the private client was “hardly” an advantage, it’s not fair to those who didn’t have access to it and were putting resources into reverse engineering. It was objectively an advantage that generated 2x-4x more revenue per node, and that is my only point.

Indeed, I remember seeing that Nodies started using geo-mesh in February:

It’s great that everything started being open-source. Grateful for the contributions Nodies has had since :slightly_smiling_face:

1 Like

Ultimately, I desire to see gateways move forward. This is under PNF’s management, and not the DAOs.

I’m very grateful for their transparency and their willingness to accept feedback. I hope my feedback has constructive merit and am excited for what is next.

Thank-you @JackALaing and @giantfrog for being open to engage :+1:


Just wanted to chime in here and say Barg Systems has never wanted to develop, manage, create or sell their own gateway nor has any plans to. Nodies asked us to be a neutral party in the initial discussions around adding additional gateways and we obliged as we were already on site in Denver and it sounded an interesting conversation to be involved in.

In the end; multiple gateways are important. This will increase competition which in turn will yield better QoS, validation, accountability and scale.


@JackALaing or @poktblade, could there be a technical spec on what these modules are that Nodies is going to be producing? I asked about this over a month ago and nothing has been released yet, so I wanted to follow-up. Basic spec questions are:

What are the modules?
What do they do?
What will their features will be?
What technologies/libraries that are expected to be used?
What tech will they interface or integrate with (HAProxy, Promethius, etc)?

I understand the Nodies won’t be doing community calls until August, but there should be initial openness on what exactly is being built, so the community knows what to plan for. So far this is the most that has been communicated:


I still have no idea what exactly is being built, and I think there can be much more precise communication :sweat_smile:

1 Like

Thanks for following up @shane . We’re still ironing out the details but here is what we’ve currently defined that Nodies will provide in return for the grant:

  1. The Gateway Operator shall:

    1.1 open source technical documentation on integrating with Pocket Network including the steps below and to allow for an agnostic language implementation:
    (i) Using an app stake to retrieve a session,
    (ii) Generate an Application Authentication Token,
    (iii) Obtain hosted dispatcher URLs or spin up own dispatcher (Full Pocket Node),
    (iv) Send a request to a dispatcher for the latest nodes in a session,
    (v) Sign a relay and submit it to one of the nodes in a session,
    (vi) Receive a response from a node, and
    (vii) Proxy it back to the user;

    1.2 open source relay client implementation of the technical documentation described in clause 1.1 above in Golang,

    1.3 Open source technical documentation on the QoS and data integrity measures intended to be implemented by the Gateway Operator, as well as the open source implementation of the QoS framework implemented by the Gateway Operator;

    1.4 provide any additional information and documents as reasonably requested by PNF to help any other gateway operator portals launch on Pocket Network;

  2. take all necessary actions to open source their Gateway Operator Portal in the manner that PNF sees fit as soon as reasonably practicable after the Effective Date;

  3. The Gateway Operator must provide details that allow POKTScan, or any other data aggregator that PNF may reasonably choose to support from time to time, to verify the number of relays that the Gateway Operator Portal is sending to the Pocket Network, including for which Chain ID, as well as all relevant details about the Gateway Operator Portal’s key operating metrics such as its reliability (uptime and correctness) and performance (latency, success rate and proportion of relays sent to Pocket Network, any altruist network, or any other destination other than the Pocket Network protocol).

To expand on the QoS and data integrity measures to be open sourced:

Node Registry - consists of the current staked nodes separated by operator identity (not address) and provides long-term historical data about an operator. This will also give us information on any potential “bad actors” in the network.

Nodies Cop - Once the servicers are registered on the Node Registry, there is no guarantee that the servicers are in healthy shape. Nodies Cop will be a construct/ideology that performs “checks” for all the nodes on the registry periodically to ensure they are healthy. In the event that one of the checks fails, Nodies Cop will pull the servicer out of the servicing rotation. Once the servicer is pulled out of the rotation, Nodies Cop will periodically test the node still, exponentially increasing the retries until the servicer is healthy again. In summary, Nodies Cop is constantly looking for infractions and pulling nodes in and out of rotation. These checks would be classified as public/secret:

  • Public - health checks that users can see in order to diagnose if their servicers are healthy.
  • Secret - health checks that only administrators of the application can see. It is only important to know that we have secret checks because we do not want malicious users to ‘spoof’ or ‘workaround’ our checks to trick us into thinking their servicer is healthy. This can be further iterated to become how the first “watcher” will be implemented within the network if successful.

Here’s a working doc detailing more: Nodies DLB <> POKT Specifications - Google Docs

Note that this is a starting point and more technical details will be shared as it progresses, through channels like Gateway Office Hours, blog posts, and forum updates. Nodies also has a point person in @giantfrog who anyone is welcome to reach out to.


Thanks for providing more details Jack :+1: From the sound of it, Nodies is being contracted to build three specific tools thus far:

  1. An SDK for using AppStakes (what is being called “Bare Bones Integration”)
  2. Node Registry (a historical record of how providers respond to RPC calls)
  3. Nodies Cop (a health/liveness checker)

Those are pretty straightforward modules. I hope that these modules are being designed in an fashion where folks can pick and choose which ones they want to use, and which ones the want to use their own modules. I know that for CC we already built our own version of #2 and #3 and could potentially make our own modules available. If Nodies’ modules are primarily designed to only work with their modules, then that would limit how others could design gateways IMO.

I do hope more POPs will be utilized as well, since these types of modules could be built others, especially ones like #2 and #3.

In Nodies presentation, they focused heavily on their gateway’s primary value proposition will be the Opinionated APIs, which indubitably requires backend indexing and catching tools. Will those tools also be open sourced as part of this initiatives? Perhaps @giantfrog may be able to provide more insight?

1 Like

Also, could you expound on this @giantfrog? The idea of Secret health checks doesn’t make sense since nodes need to know how to identity if they are not hitting a specific performance. Example, if a “Secret health check” is actually a getlogs call and their node is failing, the node runner needs to be able to know what call they failed with to troubleshoot. A heavy call like getlogs could be effected by a number of variables, but if it’s considered a “secret” then how can anyone troubleshoot or improve their nodes? How could that also be fair if a provider like Nodies knows what the Secret health checks are, but no one else does… then Nodies node’s would have preferential treatment within your own gateway.

It also seems to be hinted that Secret health checks could be part of Watchers. Has that concept been vetted by the protocol team? Again, I’m not entirely sure what Secret health checks really means, so any further explanation would be great :+1:

1 Like

This is something we will have to live with in a multi-gateway landscape. If we want gateways and apps to be permissions, we cannot control that. In the end it is the gateways reputation for fair distribution of relays what is at stake.
We plan to check how each gateway is distributing their relays and show if there is favoritism in their “secret checks” , but in the end is up to the portal user.

I would not mix watchers here, it is a completely different topic (and a large one). Checks at portal level have nothing to do with checks at protocol level (watchers). The first ones decide on the distribution of the traffic and the later ones on the amount being paid for those relays.


I get your point, but it is in the spec of what they are creating for the DAO, so I’m asking for clarification on how all will benefit from “secret” health checks, as it’s not immediately obvious what they are. They obviously have thought it through to put it in the spec, so I’m just asking for clarification.

I agree with your point. Since it was mentioned in the spec, I am wanting clarification as this would have to be a larger topic that involves the larger community.

1 Like

Thanks for this, I gave it a read and I have some comments.

Node Registry - consists of the current staked nodes separated by operator identity (not address) and provides long-term historical data about an operator. This will also give us information on any potential “bad actors” in the network.

Grouping nodes by provider name has some drawbacks:

  • Node runners like to change and/or have multiple domains to obfuscate tracking.
  • The QoS features distributions of nodes is not well defined, we have found multi-modal distributions under a given domain name.

We suggest to use a per-address registry, as it is more flexible and provides better data. We have done this in the past (when QoS data was available), we can help you with it, it is a rather large DB but is the correct way to do it (IMO).

Nodies Cop - …

I think that you read all the watcher-related data, but just in case I will add some comments here.
This highly overlaps with the watcher and you will face very similar chalenges. Specifically I recall one relates to:

… performs “checks” for all the nodes on the registry periodically …

Periodicity of checks is not wise and is easily detected, randomize this.

is constantly looking for infractions and pulling nodes in and out of rotation

Delay actions (removal) from checks, or hide actual check among other “bloat” requests to the node before kicking them out of rotation.

I’m looking forward to see your gateway up and burning some tokens!


Emphasis on why it’s a secret check! We will start by incorporating our secret checks based on our clientele’s requests and needs, and we expect that other gateway operators will do the same. For us this is a fine balance between gamers and ensuring that we are able to ensure to our clients a specific level of assurance quality and security. It is very possible that in a world of gateway verse, some gateway operators will be more strict, and some gateway operators (i.e. non-profit or loss-leader gateways) may be less strict. Nodies DLB sits in between offering Pocket as a decentralized offering to our client base and accelerating Pocket Network’s goal of multiple gateways. I believe the spec is sufficient enough in what we will actually deliver. The mechanism for how secret checks are shared and distributed is up to gateway operators to decide. It can be further iterated by future gateway operators as Pocket continues to evolve. We’re not opposed to releasing our mechanisms, but at the same time, we have to balance that with what our clients require for security since the network has no effective slashing mechanism at the time.


I’m personally not a fan of the per-node system we have in the Pocket ecosystem right now. It is often a sign of confusion about what the number of node counts actually means when in actuality, there is only a certain amount of blockchain nodes operating behind them. Ethereum is also going through a somewhat similar issue where they’re proposing to allow for stacking vertically instead of horizontally for nodes since many nodes map back to a single identity, anyways. That being said, it’s a pretty minor detail for us right now and can always be iterated on.

There have been many ideas we’ve floated around (i.e: an open dataset from us of ranked nodes that node operators can use as a baseline then build and train their own model, but this is likely not to make it into the initial bootstrapping budget, perhaps as a continuation and evolution of the initial gateway-verse). Any collaboration we can have with the best data science team in the POKT ecosystem sounds great :sunglasses:, and just shows the potential of what the world of multiple gateway operators would look like. I’m really excited to start forth on what we can do for the ecosystem and the opportunities it opens up for others.

An update on our end:

I’m happy to say we reached terms with PNF, and the last action item is a formal execution on contractual agreements in regards to our deliverables. We will finish this today and will have news to share with the community after our roadmap planning (1-2 weeks), and our first payment received for this budget will be started on the 1st of September :partying_face:


I know how much time these forum posts take from the initial post to replying to and answering all of the questions.

Shout-out to everyone involved to get this over the finish line.

Send it!


A brief but crucial update in spirit of open data and transparency!

  • POKT <> Nodies Integration Public Tracker can be found here , this tracker outlines all the tickets we are currently working on. This dashboard can singlehandedly be used to track all our progress.
  • POKT MVP Integration date is preliminary slated for October 31st & we’re currently a week ahead in progress.
  • User onboarding guidelines and baseline benchmarks published on our docs. We will have material for POKT before integration is complete.
  • website was revamped both from a code perspective, observability, and UI enhancements. We are working towards another addition to highlight Pocket, which is staged on
  • @Jorge_POKTscan is graciously working on Testnet implementation for Poktscan so we can leverage it for gateway testing. (:fire: kudos to Poktscan team for taking this on). I cannot stress the importance of testnets.
  • Marketing for multi-gateway awareness is a work in progress in collaboration with @Adrienne (PNF) and our BD team!

4 of us, including myself, will be in Permissionless II, so if you will be attending and want to connect - please on Twitter/X!

Track our progression in our discord, public project tracker, through DM’s, or our socials.


Thanks for the update.

Can you explain why the QoS framework module will only “partially implement the ‘Cop’ model”? Why does the QoS module not include the whole Node Registry and Nodies Cop functionality that Nodies is currently building?

I understand that the you will be providing “technical documentation on how we’re currently identifying good actors”, but since only part of the code will be open source, this means that future gateways will have to spend their time and resources reverse engineering all the parts that aren’t open source.

Thanks in advance :+1:


As Dermot recently shared in PNF’s ambitions update, our Gateway-verse objectives for Q2 2024 are as follows:

  1. Onboard two new high-potential gateway operators, each capable of delivering 1B+ daily relays to the network
  2. Build a pipeline of 5+ additional high-potential gateway operators, each capable of sending 500m+ daily relays to the network - to be onboarded at the launch of Shannon
  3. All new gateway operators (excluding Grove) sending an aggregate of 5B+ daily relays to the network

With these ambitions laid out, and some Gateway-verse milestones recently hit, it’s time for an overdue update on where we’re at with Gateway-verse. Moving forward we’ll be working to uphold a higher standard of update frequency from both PNF and Nodies.

This update outlines 1) what has been worked on so far in the Gateway-verse and, 2) how we’ll work towards achieving our Q2 2024 objectives.

What has been worked on so far:

  • Onboarding Nodies: PNF defined a standard gateway operator agreement, which will enable wider onboarding of other gateways, as well as a grant agreement with Nodies. The grant began September 1st.
  • App Stake Management: PNF has been working with Grove to free up app stakes for other gateways. Nodies now has enough app stakes to do more than 400M requests/hr, matched with 624 nodes every hour.
  • Nodies Integration: Nodies completed their integration with POKT on October 31st, which included
    • Optimized Golang Client: Nodies has been focused on extracting as much performance as possible for their gateway, building their own custom minimal Golang client to their relay specification. They’re experimenting with FastJSON) and FastHTTP) to optimize on latency rather than use the standard Golang packages. This has led to 2-4X performance boosts in the relay process, which should translate to more efficient gateways in the long-run as these optimizations will be embedded in the OSS Gateway Stack and will carry forward into Shannon. We’re expecting Blade to share more detail in this thread later this week.
    • Dispatcher Management: they’ve set up their own dedicated Pocket dispatchers to enable better scaling of traffic through to POKT
    • Load Balancer between Nodies and POKT: Nodies has the ability to serve endpoints from their own nodes as a backup, so they have built an internal load balancer to enable dynamically shifting between their own nodes and POKT
    • QoS Checks: they’ve implemented basic QoS checks within their own systems, including a height checker, a latency checker, and a simple node registry
    • Observability: they’ve implemented observability on all their backend and frontend systems to allow them to determine how a request is performing in terms of success rates, requests per second, per chain, etc. This allows them to be alerted whenever there is an issue and to ensure a high standard of QoS. They’ve migrated their observability away from postgres and into a more purpose-built time-series database using Prometheus, Influx DB, Posthog, and Grafana. These are all open source technologies that anyone can integrate and will lay the foundation for the Observability Service within the OSS Gateway Stack, as well as upgrades to the data provided to customers in their frontend.
  • Public Endpoints: Nodies completed their integration with POKT in time to take over the role of maintaining public endpoints from Grove. These are now serving just shy of 100M daily relays. Thanks to their work on building an efficient baremetal gateway, we’ve observed a 60% reduction in costs to maintain the public endpoints.
  • Gateway Dashboard: POKTscan updated their dashboard to display per-gateway relay performance. Here you can view gateway relays per-chain, in time series, and as a % distribution.

How we’ll work towards achieving our Q2 2024 objectives:

  • More Frequent Updates: Both PNF and Nodies will be providing more frequent updates here in the forum and on community calls. For the latter, the Node Runner Office Hours have evolved into Ecosystem Office Hours in part to make space for gateways to be discussed on those calls.
  • Optimizing the Integration: Nodies has been paying close attention to their POKT integration, tuning their performance with POKT using real traffic, adjusting their QoS checks, and ramping up the traffic that is coming through the network.
  • Building the OSS Gateway Stack: The vision for the gateway stack has evolved from a barebones integration and QoS framework, into a modular architecture powered by GRPC, with simple Docker deployments and the flexibility for developers to choose only the services they need and/or plug-in other services. Now that Nodies has completed their vertical integration and are sending traffic through POKT, they are focusing on modularizing it. The repo in which this will be developed should be shared by Blade this week, starting with the optimized Golang client that we described above. Everyone will be able to follow the progress in GitHub and we expect the stack to be ready for onboarding by early adopters by late December, then completed (RC ready) in January. You can read more about Nodies’ vision for the stack here.
  • Headhunting Gateways: PNF has begun deploying a headhunting GTM strategy targeted at ideal gateway personas, meeting with high-value gateways, understanding their needs, and establishing an onboarding pipeline with a view to onboarding them starting in January. Please get in touch directly with either myself or Dermot if you can introduce any established RPC businesses capable of sending 1B+ daily relays through to the network that might be interested in building a gateway on POKT Network.
  • Public Applications: PNF will launch a public application process for becoming a gateway operator, so that anyone interested in building a gateway can be considered for the opportunity. If you don’t want to wait for the application process, feel free to get in touch directly with either myself or Dermot.
  • Preparing to Onboard Gateways: PNF is preparing an onboarding/support package for gateway operators, including one-pagers, ecosystem support offerings, public docs, and economic models (cost, breakeven).
  • Consistent Gateway Marketing: PNF will establish gateway marketing guidelines for use by all existing and new gateways, including “Powered by POKT” assets and requirements to mention POKT in materials/promotions.
  • Sockets/POPs: PNF also plans to use the ERA Allocation to create more grant opportunities for parallel community projects that complement Nodies’ OSS Gateway Stack work (rather than duplicating it), such as QoS monitoring solutions.

I’m excited to see the gateway space moving forward! OS-Gateway looks like a great start to this budding ecosystem.

All these are great accomplishments, however they don’t translate into anything tangible for the community or other gateways, since these are only relevant to the setting up of Nodies own gateway.

I understand that more modules are currently being built to address some of these areas, but it’s not clear what the full stack of a gateway will look like with these modules in a production environment. Very basic info has been shared, but none of that is actionable or detailed enough to fully understand how a production stack is suppose to work.

@poktblade could you provide more insight? My starter questions are similar to what was asked in July:

  1. What is the flow of a relay in your architecture?

  2. How does this stack lay out? (Example: Does OS-Gateway go right behind your firewall, or does it go through an LB, or call filter, etc?)

  3. Where do future modules expect to be placed in your stack? (Example: Will the QoS module connect into OS-Gateway, or does is interface directly with an LB like HAPoxy?)

  4. What dependencies will these modules have?

  5. What tech will these modules interface or integrate with (HAProxy, Promethius, etc)?

@JackALaing said last month :point_down:


I replied that without code being openly developed and without technical details on what how this stack is supposed to work, contributors like myself cannot get involved. I’m trying to understand the envisioned architecture and see where folks like myself can get involved.

Open-sourcing OS-Gateway is an awesome start, and from my understanding all future development will be in public repos, which will be great as well, but folks like myself need specifics so that we can participate.

Today it was mentioned that a forum update is in the works from Nodies, so I wanted to get my questions out ASAP :+1:


Thanks @JackALaing for this recap. To add more to the Gateway Stack:

The Nodies Gateway Stack aims to streamline the integration of applications with the POKT Network, reducing the complexity associated with directly interfacing with the protocol with an all-inclusive stack that any developer can contribute to. This allows for mass adoption across multiple personas such as application developers, existing centralized RPC platforms, future gateway operators, and more.

Whenever proposing and pioneering the pathway to multiple gateways and its benefits to the Foundation, they opened up with open arms to enable us. Whenever we wrote the initial technical specification, the reason why it felt “ambiguous” to some was to prevent being locked into a granular statement of work given that the potential scope of this project is extremely large and to reduce the friction of unnecessary bureaucracy. The specifications were structured in such a way that the ecosystem would always benefit. Our team continues to push the limits of what we can actually do for the ecosystem based on our expertise and experience with the protocol and as a RPC provider, ultimately leading us to the gateway stack.

So what exactly is the gateway stack right now? The gateway stack kickstarts off as a light-weight process that enables developers of all kinds to be able to interact with the protocol and engage with 50+ blockchains without the need of storing terabytes of data, require heavy computational power, nor understand the POKT protocol specifications with a simple docker compose file. The rhetorical question that we pose to future actors who want to maintain a blockchain node is: Why spin up an Ethereum node and maintain it yourself whenever you can just leverage POKT natively using the Gateway stack? Afterall, using POKT would require a fraction of the required resources and technical staffing.

We envision that the stack will be used as a foundation for their entirety of the ecosystem to continue to build on top of such as:

  1. Building their own frontends and extending their backend to include POKT by using the Gateway stack for their own SaaS business
  2. Using POKT as a hyperscaler whenever they need more computational power or access to more blockchains (sticking the process into their LB rotation)
  3. Using POKT as a backend as a failover whenever their centralized nodes goes down (sticking the process into their LB rotation)

The use cases are limitless and we expect that over time, community contributions into the gateway stack will enable some of the aforementioned use cases natively. Below is an example persona we’ve created to illustrate how existing RPC providers can use POKT through the Gateway Stack.

Next steps

Our plans for this December and January to transition from alpha to RC ready ordered by priority:

  1. Enable QoS checks that will allow Pocket responses to return with a 99% success rate.
  2. Enable Observability on the gateway server to measure success, error, session dispatching rates (Allow for Prometheus to scrape these metrics)
  3. Create E2E test coverage that allows for a reliable release process.
  4. Create documentation and architectural diagrams to elaborate on the existing state of the stack
  5. Assist the Foundation to help onboard more gateway operators

Our goal is that by the end of the 6 month period, the fundamentals of the Gateway stack are sound, and there is a sustainable and realistic way for many more community contributions down the road. Our team will continue to own the direction and vision of the project to prevent any unnecessary thrashing. However, we welcome any discourse within reasonable bounds and our roadmap will continue to be shaped by future gateway operators who will onboard the network.

Please note that the actual deliverables that we are committing to has not changed, the Nodies Gateway Stack exists to house these deliverables but as well set a long-standing vision for future contributions. The end state of the Gateway stack will continue to evolve, however the next two months are now granularly set to baseline the expectations.

Objectively taking a step back and looking at the larger picture…

To dive deeper into our accomplishments, we frame it in respect to the objectives outlined in the OP and the timeline which is about the hafl way mark of the total engagement as far:

There was a clear need to define processes for future gateway operators (i.e gateway docs, relay specification, app stakes distribution mechanism, identifying the need to securely rotating the app stakes, contractual agreements, etc). We have been actively working with PNF to help remove these logistical barriers. As a result, these processes will be then distributed to the next 2-5 gateway operators that will be onboarding the network in the upcoming quarters.

:white_check_mark: On track

The grant’s purpose allowed us to build both inwards and outwards. Both of these characteristics go hand to hand, and the POKT ecosystem does benefit from both. It was always our intention that would directly integrate with the protocol as a first step and to focus time and energy on building out BD/Sales for POKT. Then we would fan that out into a more open source-project to help further gateway operators onboard.

We are now powering anywhere from 50m to 80m POKT traffic a day at a 60% cost reduction in comparsion to previous gateway offerings. This only cements that multiple gateway operators will be a net add for the ecosystem. Furthermore, this is a huge step forward for the POKT ecosystem because:

  1. POKT’s traffic is now more resilient than ever. Gateways now are operating on multiple different infrastructure stacks leading to a further resilient network as gateways are spread across multiple PoPs ranging from GCP to Bare Metal.

  2. By directly integrating, this has led us to learn more about what it actually takes to become a gateway operator and ultimately strategize on what we would need to solve for in order to reduce the friction in gateways. Ultimately, this was one of the major contributing factors to our open source gateway stack vision and the actual OS Gateway stack.

:white_check_mark: On track

By directly integrating with the protocol and serving real world traffic, this has given us sufficient information on more QoS checks we would have to perform, and where we can optimize. This has already been reflected in our cost savings for the Network and why the gateway stack is so focused on performance and effiency. Furthermore, the prortocol team has already reached out to us to talk more about Gateways in general and where the OS Gateway stack fits in a post Shannon world.

:white_check_mark: On track

We already have our platform hooked into the network powering public RPCs. By the end of Q2, the foundation has set a goal of 2-5 gateway operators being able to also power POKT RPC. There have been numerous queries behind the scenes from both ecosystem members, node runners, and even centralized providers on leveraging our gateway stack. The rate that we are going, there will be plenty of actors who can potentially be tapping into the network for RPC on the demand side, ultimately cementing that the POKT Protocol is the only decentralized RPC network on both the supply and demand ends.

:white_check_mark: On track

Closing Statement

Our recent efforts have made significant strides in achieving the outlined objectives within the past ~3 months. From removing logistical barriers for future gateway operators to supporting a second-mover gateway, learning about gateway operations, fostering innovations in technology, and diversifying revenue sources, Our integration and contributions have demonstrated tangible progress. The achievements so far continue to pave the future for a multi-gateway ecosystem that enhances the overall resilience, efficiency, and value of the Pocket Network :rocket:


@poktblade This is a perfect way to sum up how the past 6 months have been from someone on the outside :point_down:

From what I got from the initial announcement, I don’t believe this was supposed to be a Nodies only “direction and vision”. I saw it as the DAO bootstraps Nodies to be a steward of leading it’s development, but once it actually began, it has morphed into a Nodies only show.

From an outsider’s perspective, who’s been trying to get a handle for 6 months now on the technical aspects of the stack being built with DAO resources, there is the same level of exclusivity and siloing that reflect what many criticized when POKT was under a single company. Why do we want to repeat history? While I’m glad that the protocol team is getting all the technical information, this “picking and choosing” who gets to understand your actual gateway or tech stack is wrong IMO.

If I’m to take your comment at face value, your stack is now effecting protocol design, yet it’s architecture is still secretive? Though PNF has exclusively paid Nodies to built it, I don’t believe it’s design or architecture should be deemed exclusive for only some teams.

Overshadowing The Good

I’ve already expressed before that I believe you do solid work, and the code from my initial review of the OS Gateway was positive… but for me, it is overshadowed by the fact that any technical questions have been ignored (at least from the public) and we are now being told that we will only get transparency at the tail end… after it’s been built and the funds have been mostly spent.

On one hand, I should be excited that more gateways could be on-boarded next quarter… but on the other hand, the technical side is an exclusive club, and any attempt to make it inclusive (through questions) are ignored, while the exclusivity is being framed through a lens of justification. All this is worth addressing IMO, instead of pushing down the road to address in some other future intuitive that adopts a similar approach. Large 6 month initiatives (which represented nearly 1/4 of an ERA’s budget) should not be treated as some small initiative that are finished in a few weeks.

@poktblade, I think your stack probably has a straightforward architecture, and the gateway stack vision likely makes absolute sense… so I have no idea it can only be selectively shared. Whatever is being shared with select teams (like the protocol team) should be shared openly IMO.

Looking forward to when the gateway ecosystem doesn’t have gatekeeping :+1:

I’d suggest making your #4 priority, your #1 priority, as that is more in-line with the initial spirit of this initiative. Since it already exists (deriving from your comment), why not just communicate it now, instead of at the tail end of this initiative?

There seems to be this framing that any meaningful technical sharing is burdening of progress, and it should only happen at the end… but that is not a precedent that should be re-introduced into POKT, IMO.

@poktblade how many of these relays are from the public endpoints that are being paid for by PNF? Do we have a track on how much of your gateway burn is from the DAO resources vs other traffic?

Critical Disposition

You’ve shared some impressive stats about your gateway and it sounds like a good overall vision, but I struggle to get excited about something that has been shrouded in exclusivity. I’ll probably lean toward a critical disposition until this is no longer exclusively a “Nodies + select others” initiative, which I don’t believe follows PNF’s original vision shared in June.