Old School Runescape launches servers in Brazil

Brazilian Game Worlds Are Live! Brazilian Game Worlds Are Live! 09 April 2026 Brazilian game worlds are here! As of today, we have officially opened four game worlds based in Brazil! This means better ping for players in the Latin America region, making high-intensity gameplay much more feasible. Take down Doom of Mokhaiotl, join your clannies on a high-stakes raid or get grinding your Thieving, all with less interruption. While our new servers are based in Latin America, it is a huge region and players in some locations may find they still get the best latency from our US worlds. If you can’t see the image above, click here ! You'll now be able to access the following worlds: 692 - F2P 693 - Members 694 - Members 695 - Members Find the new Brazilian game worlds in your world list, now! Releasing this work a week ahead of Leagues means we're able to open two more dedicated Leagues Brazilian game worlds for the launch of Leagues VI! Plus, with the infrastructure in place, we have the flexibility to add more if we need to. The journey to Brazil wasn't a simple one, so we caught up with the infrastructure team for the full low down. Game worlds based in Brazil are coming to AWS . That sounds simple enough, but getting there meant changing a fair bit about how we run Old School in the cloud and how our network fits together. The infrastructure J-Mods are back with another Spudworks blog. This time, we're talking about Brazilian game worlds, cloud hosting, and why doing this in 2026 looks very different from doing it 10 years ago. As usual, this is the technical cut. We wanted to give a bit more insight into what had to change behind the scenes to make Brazilian game worlds in AWS a real thing rather than just a nice idea. Regional hosting: A history Since this is Spudworks, let's take a quick trip down git history lane . If you've been around long enough, you might remember that we used to run an extra United States East region for RuneScape: If you can’t see the image above, click here ! United States East 2 - Miami Beyond that, back in 2014, we had small presences in Incheon, Korea and Cape Town, South Africa with our own servers, including RuneScape World 51 and RuneScape 106 . If you can’t see the image above, click here ! If you can’t see the image above, click here ! If you can’t see the image above, click here ! If you can’t see the image above, click here ! For higher resolution, click on the image to open it separately and zoom in to your heart's content! Ultimately, we decided not to continue with these locations, mainly due to the cost compared to the lack of performance gains we were seeing. They proved to be an interesting experiment though, and we definitely learnt from it, including when we accidentally routed all traffic via Cape Town! Why Brazil, why now? For a long time, Miami was our answer for serving Latin America better. It was closer to South America than North Virginia, it was simpler for us to run, and at the time it made more sense than trying to set up our own estate directly in Brazil. We have looked into the feasibility of setting up shop in Brazil regularly over the years, but the problem was that in the 2010s, hosting there directly was a much bigger ask: Hardware, facilities, support, operational overhead, and legal complexity all looked a lot harder than they do today. So Miami remained the compromise. While we know it did help, we also know it only went so far. For a lot of players, the latency improvement versus North Virginia was modest rather than dramatic, and over time that region made less and less sense as a long-term answer. The hardware was getting old, leading to an increased hardware failure rate, and really it was only running RuneScape and Deadman tournaments for Old School due to limited capacity. So, on February 14th 2023, we collapsed back down to a single US East region in North Virginia. We refreshed all of our hardware and expanded our US East data centre to be the largest of them all. This brought considerably faster hardware and network equipment, along with improved architecture for world deployments, making for much more reliable server ticks. But, the desire to better serve our South American players never went away. We continued to investigate ways to bring Old School hosting to the region and save players from making the pan-American trek to our Virginian potato farms. Last year, in July 2025, we ran a survey to gauge where you would prefer us to start. The results from this survey solidified our decision to head to Brazil as our first port of call. If you can’t see the image above, click here ! The Journey to Brazil One of the reasons Brazilian worlds are possible now is that we have taken the time to experiment and investigate thoroughly. In March 2023 , our team took on exploring the ability to add a Brazilian cloud infrastructure to our production environment as part of a Game Jam. From there, this stopped being a theoretical discussion and started becoming something we could actually do. If you can’t see the image above, click here ! In October 2024 , we ran Old School worlds 596 and 597 on AWS in Australia as our first production prototype, with players being able to access and use the worlds. We initially compared tick performance between our existing worlds, and x86 and ARM (graviton) EC2 instances on AWS. We identified some performance gains on ARM compared to x86, and the engine and infrastructure changes which would be required to run this setup. Between that point and where we are now, two important things happened based on what we learnt running the Australian prototype: We made major changes to Jagex's global network infrastructure; and Project Zanaris delivered some engine improvements which helped Old School worlds run on cloud infrastructure more natively. Since November 2025 , Old School worlds 596 and 597 have been running on AWS in US East as salvaging worlds on newer architecture, improved by the learnings of what didn't work with the previous prototype. We now have good data on what different game modes and activities cost to run in AWS, and the performance gains related to it. That matters because Brazil-on-AWS is not us jumping straight from an old hosting model into a brand-new region and hoping for the best. We've been constantly chipping away at the pieces needed to make cloud-hosted Old School worlds feel like part of the normal estate. It's not flashy, and it doesn't make the patch notes, but it is this kind of groundwork which makes new regions possible. What changed in the network? If you remember our blog post back in August 2025 , we shared that we had fully completed setting up our infrastructure with hybrid connectivity globally. Here's a refresher on the handy diagram we shared to help illustrate the work, and how it's changed since then. If you can’t see the image above, click here ! The backbone was originally tied together by a VPLS, and is now fully replaced with Direct Connect with SiteLink . So, what does that mean? Well, VPLS (Virtual Private LAN Service) essentially allows you to set up a network that acts as one big network. This is what our game worlds and backend services constantly talk to each other through. We would connect each of our data centers onto a dedicated provider VPLS network. Meaning, if we ever came under a DDoS attack, our services could still communicate and, most importantly, your save game would be preserved without interruption. However, we wanted game worlds on AWS and on-premise, so our backend services and game worlds need to be able to reliably communicate no matter where they're located. This is where Direct Connect with SiteLink comes into play, letting us achieve the same outcome as when we were using our VPLS to advertise our networks to the other data centers via BGP. BGP (border gateway protocol) is a system that tells different networks "here is the best way to reach me", so data knows where to go across large connected networks. We set up the AWS replacement with SiteLink enabled, as it relies on BGP to exchange network information between connected sites and our AWS network's VPC. It also acts as a data-plane capability that allows traffic to flow directly between our sites using AWS as Jagex's global private backbone, all while giving us the ability to connect to our game worlds and backend services on AWS at scale. Now, instead of treating cloud regions like distant add-ons, we can connect them into the wider estate in a way that is much more natural operationally. With this work complete, it is considerably faster to spin up new regions, in principle. However, it doesn't mean your favourite region is necessarily "Coming Soon ™", cost, legal considerations, and total player benefit all affect which regions we may provide. AWS Cloud WAN To be able to maintain a global network on-premise and on cloud with a fairly large AWS estate, we make use of AWS Cloud WAN . In simple terms, Cloud WAN acts as a centrally managed global network hub within AWS. It allows us to define how all of our regions and networks should connect to each other in one place, rather than configuring each connection individually. Under the hood it still uses familiar concepts like routing and segmentation, but wraps them in a managed service that handles the heavy lifting of keeping everything connected and consistent. Cloud WAN gives us a more structured and scalable way to organise that connectivity within AWS itself. Instead of thinking in terms of point-to-point links or manually managed routing between regions, we can treat the AWS side of our network as a cohesive backbone with centrally defined policies. This becomes particularly important as we scale. As we add more regions or services, we don’t need to re-architect how everything connects each time. Instead, we extend the existing network model in a predictable way. It reduces operational overhead and makes the network easier to understand and manage. However, this does not mean every region is identical, each will have its own intricacies to navigate. When setting up for our Brazilian game worlds, we had to account for the fact that Sao Paulo (sa-east-1) is one of the few regions that does not support Cloud WAN yet . So, while many of our newer models made Brazilian game worlds easier, we still had to do extra work with AWS Transit Gateway in that region. While the cloud does make some things easier, it took the hard work we've done over the past few years to make even the most complex cases now more practical in a way they weren't before. Why does that make Brazil viable? Back when Miami was chosen, a Brazilian game world deployment would have looked much more like a one-off regional build with a higher operational cost and less integration into the rest of the platform. As well as being unsustainable, it would have provided a second-class experience for users in this region, which we weren't willing to accept. Today, the picture is different. We now have: A true hybrid network model Practical experience running Old School worlds in AWS Engine changes that make cloud-hosted worlds behave more naturally than they did in earlier attempts Put all of that together and Brazilian game worlds stop looking like an isolated special case and start looking like something we can support properly. What has really changed? If we boil the whole thing down, Brazilian game worlds in AWS are possible now because several different threads have finally lined up: We stopped leaning on older regional patterns that were only partial solutions. We have a process now for cloud-hosted Old School worlds done in stages rather than in one big leap. The network moved from a more traditional VPLS-connected estate to a hybrid AWS-connected model. Some of the work from Project Zanaris helped Old School run in the cloud more naturally. We did the extra work needed to make Brazilian game worlds viable, even without AWS Cloud WAN in São Paulo. The Brazilian game worlds announcement is the main headline, but behind the headline there has been a considerable amount of iterative network, cloud, platform, and engine work which made it possible to move Brazilian game worlds in AWS from whiteboard to reality. If you'd like to see more of these deeper dives into how we run the infrastructure behind Old School, let us know! You can also discuss this update on the 2007Scape subreddit , the Steam forums , or the community-led OSRS Discord in the #gameupdate channel. For more info on the above content, check out the official Old School Wiki . Mods Adad, Bash, Boko, Cky, Drax, Haydon, Ibex, Kraken, Maniac, M0iqp, Qwert, Roman, Titus, Vallcore, Vxp ... and🐹 The Infrastructure J-Mods