Potentially obvious copyright claim for the title of this post aside, I've been working on upgrading api.deathwing.me lately.
For some background information, api.deathwing.me handles over 3000 active connections every second with tens of millions of requests each day and honestly, it is safe to assume that it is the second most popular API node after hive.blog -- with monthly bandwidth usage crossing 10TB mark in the last few months, I decided some changes to improve stability were due.
Recently, I was able to get into Top 20, this allows me to invest further into my Hive infrastructure setup. So as a result, I've gone ahead and added/changed a few things in my API node.
So... what changed?
First of all, as you may or may not know, api.deathwing.me is hosted on a server equipped with Ryzen 9 5950X and 128GB DDR4 ECC RAM, in other words, it's pretty much top of the line when it comes to technical specifications of a server build at the time of posting this.
To improve performance (as sudden load spikes were able to take down my node since I do not, as opposed to other node providers, rate limit total calls) and ensure the node is stable in an event of a failure in the rack/datacenter/hardware itself, there are now two servers, instead of one.
These two servers are connected to a load-balancer that utilizes Cloudflare for global network routing, which in turn, hopefully, means faster "resolve" times as well as faster routes to the API node in itself, in short, actions and API calls feels "snappier" and "faster" overall.
TL:DR
api.deathwing.me is now on a global network with two top of the line servers to ensure uptime (so even if one datacenter burns down, the other one is fine) and high performance (since calls are shared between two beefy servers, it's not enough to bog them down) to ensure better usability for the end-users, which means you, the person reading this post.
Stats
It hasn't been super long since I've adjusted the DNS, so the stats below might be less than what some people expect as DNS propagation and caching for users might not be up-to-date and data should start to change in the next few days or weeks. But what you see below is the last 24 hours of data from api.deathwing.me -- over 10.65k unique IP addresses and close to 40M requests. For the record, as I write this post, there are 4500+ active connections to the load balancer.
The future?
In the future, as the chip shortages alleviate and prices start coming down even further, I am planning to get even more servers to ensure full global coverage, however, this does cost quite a lot, so I might revitalize my DHF proposal in the future when I have time for it.
Sounds awesome! How can I support you?
If you enjoy and utilize my projects on Hive (of which there are quite a few) you can simply vote me as a witness on the ecosystem.
To do so, you can visit here to vote with Hivesigner or Hive Keychain.
That's it, been so long since I posted anything.
Can you make a bridge from hbd to polygon?
This post has been manually curated by @bhattg from Indiaunited community. Join us on our Discord Server.
Do you know that you can earn a passive income by delegating to @indiaunited. We share 100 % of the curation rewards with the delegators.
Here are some handy links for delegations: 100HP, 250HP, 500HP, 1000HP.
Read our latest announcement post to get more information.
Please contribute to the community by upvoting this comment and posts made by @indiaunited.
Your content has been voted as a part of Encouragement program. Keep up the good work!
Use Ecency daily to boost your growth on platform!
Support Ecency
Vote for new Proposal
Delegate HP and earn more
Congratulations on making the Top 20. Your efforts to the entire ecosystem are well worthy of that position.
Question: Is the traffic being split between the two servers or is the second simply a back up to the first? If I am reading this correctly, the load is divided up between the two?
Thank you so much!
To answer your question: In regular operation, the load is divided up between the two. However, the servers also act as a failsafe for each other, so in an event where one of the two servers suddenly decide to not show up for work, the other one picks up the load. This happens automagically before the user even realizes there might be something wrong.
Congratulations on reaching top 20. People like you should definitely be encouraged and i would even call you one of the strongest pillar of Hive for running such a powerful node without rate limiting. I wish you reach top 10 soon. Cheers!
I came to this post because I liked the reference to Daft Punk and I see good news. Congrats on making it into the top 20 man!
good job bro
good 😄