At Geocode.Farm, we’re always striving to provide the best service to our customers. Recently, we've experienced some downtime, latency, and other issues that we've been working hard to address. We understand how crucial reliability and speed are, and we’re making a major change to enhance both.
We’re moving away from our old-fashioned self-hosted datacenters and upgrading to a more reliable, innovative solution. But here’s the twist – we’re not going the typical "cloud" route, and we’re certainly not relying solely on CDN services like Cloudflare. While Cloudflare helps, it’s just a pass-through for our API, which requires real-time backend access for user authentication, query counting, and more. So, we needed something better.
Instead of following the crowd, we’ve decided to get creative. We’re leveraging our own IPs and implementing an Anycast network using Vultr to solve our performance issues. What does that mean for you? Faster, more reliable access to our API, no matter where you are in the world. And no downtime ever again... If one location is for some reason down, it'll just fail-over to the next closest - seamlessly.
Key Locations for a Better Experience: We’re starting by opening servers in 5 strategic locations to ensure low latency, fast load times, and optimal performance across the globe:
-
Amsterdam, NL – To serve Europe
-
Los Angeles, CA – To serve the Americas
-
Sydney, AU – To serve Oceania
-
Singapore – To serve Asia
-
Johannesburg, ZA – To serve Africa
These locations will help us cover every continent with decent low latency and fast processing times, while distributing the load across multiple sites to avoid congestion.
What’s Coming Next: We’re not stopping here. We’re planning to add even more locations with the goal of getting our global full load time (from the time of the query to processing and return) under 100ms. Here are the future locations we’re aiming to add:
-
Delhi, IN – To serve India and the Middle East
-
Seoul, KR – To serve Far-East Russia and the rest of Asia
-
São Paulo, BR – To serve Central and South America
-
Miami, FL – To serve Eastern USA, Canada, and Mexico
-
Réunion Island – To serve the Indian Ocean region
-
Hawaii, US – To serve the Pacific Islands
These improvements are all part of our commitment to providing a fast, reliable geocoding service no matter where you are in the world. We’re excited about these changes and are confident that they’ll make a big difference for our users.
Why We Chose Vultr for Our Anycast Network: To power our new infrastructure, we’ve partnered with Vultr, a provider that gives us the flexibility and reliability we need. With Vultr’s VPS line, we can scale our resources to meet our needs, and their NVMe Block Storage allows us to expand our storage capacity quickly and efficiently as we grow. Plus, Vultr's BGP availability in all of their locations allows us to use a single IP address across all locations. This means we can keep the service behind Cloudflare, avoiding the need for DNS round-robin or geoDNS solutions, which are becoming less effective as anycast public DNS providers become the norm. This approach ensures a seamless experience for our users, no matter where they are accessing our service.
Expected Downtime During Transition: As we roll out these changes over the next week, we want to let you know that there may be some minimal downtime while we transition to the new infrastructure. Once the new servers are up and all data has been synced over, we’ll need to turn off our old servers to perform a final sync of account data. This process will ensure no changes are made during the sync and that all data is consistent before we update Cloudflare to point to the new anycasted IP.
The downtime should be minimal — likely less than an hour — but we wanted to give you a heads-up so you can plan accordingly. Most of our data (like geo databases) is read-only and isn’t frequently updated, so we’ll be able to sync that live without issues. However, the user account portion needs to be synced offline to maintain consistency during the switch-over. After the sync, all of our databases across all servers will be set up to sync in near-real-time. This means, for example, that if you process a query in the USA and another user does the same in the UK, both query counts will be logged simultaneously across both servers.
We’ll notify everyone via email at least a few hours before the downtime, so you’ll have time to adjust. Again, we expect the downtime to be brief and the transition to be smooth.
Why Vultr? If you’re interested in improving your own infrastructure, we highly recommend Vultr for its performance and scalability. With their VPS solutions, NVMe storage, and BGP-enabled locations, they offer everything we need to make these upgrades possible. And as a bonus, if you sign up through this referral link, you’ll receive $300 in credit to use over the next 30 days. It’s a great way to try out their services and see the benefits firsthand!
These improvements are just the beginning, and we’re looking forward to delivering even better performance for our users. Thank you for your patience as we make these exciting upgrades! Stay tuned for more announcements in the future.
« Back