To deal with the ever growing load, and general instability of Steem server software, we've made some changes to our RPC node.

Our node is one of the most popular nodes ever since Steemit retired their normal RPC node, and now only host a Jussi node (which is not compatible with some Steem libraries due to the lack of websockets). However, having just one node which can go down means that many applications break whenever our RPC node is overloaded or is replaying.

In this post we discuss how we're fixing this, allowing us to handle the increasing load.

![thumbnail2_thumbnail.png](https://steemitimages.com/DQmYg75g3rFgBNV34zqDTRBV2S4P1epDLww3tgfenqiwnE5/thumbnail2_thumbnail.png)

# What's new?

`steemd.privex.io` is no longer behind Cloudflare. This allows our rate limiting to work better, and may increase performance.

@kryogenic has fine tuned our nginx config, allowing for smooth failovers to alternative RPC nodes in the event ours is down, this allows our RPC node to be one of the most reliable, as it will automatically failover to other people's RPC nodes when ours is under heavy load, or is offline for replays/upgrades.

We currently use the following RPC nodes as backups. Please email support [at] privex [dot] io with the subject "RPC Addition" if you would like us to add your node. It MUST be a full node, with all plugins enabled, and MUST have working SSL.

```
server rpc.buildteam.io:443
server steemd.pevo.science:443
server gtg.steem.house:8090
server steemd.minnowsupportproject.org:443
```

We do NOT use Jussi nodes, as some applications are not compatible with Jussi. All nodes are standard STEEM RPC nodes, with websocket support.

# Future Plans

We'll be adding better health checks to the load balancing system, allowing it to automatically remove unstable nodes, prioritise faster nodes, and remove nodes which are out of sync. This will allow for our service to be faster and more reliable, so that your applications work flawlessly with our server without worrying about having to change the RPC node when something goes down.


# Our server

Our RPC server is located in Germany. The specs are as follows:

---|---
---|---
**RAM**|512gb DDR4 ECC
**CPU**|AMD Ryzen 24-core 48-thread
**Disk**|2x 950GB NVME drives in RAID 0
**Network**|1GBPS network from Germany

Our server has Websocket support, SSL, as well as IPv6 support :) To top it off, if you are on the Privex Germany network, our RPC server will not add to your bandwidth cap - network requests to our server within our German region are free.


# Direct Access (IMPORTANT FOR DEVS)

As we know, some applications do not play nicely when load balanced due to variations in plugin numbers, and minor config differences.

If you require direct access to the Privex node, please use `https://direct.steemd.privex.io`. **PLEASE do not use this unless it is absolutely necessary.**

At the current moment in time, it is offline as it is replaying, and will be back online within 48 hours.

# Block Indexing

If you do any sort-of block indexing (`get_block`), e.g. for a database, please DO NOT use a public RPC server.

It is possible to set up a local RPC server for block indexing with very little RAM, you do not need a full RPC node to do block indexing. Simply set up a standard seed node and you will be able to use get_block without taxing the network.

# Thank you!

Please vote for our witness to help us fund more infrastructure like this for Steem projects. To vote for @privex, go to https://steemit.com/~witnesses and enter in the box at the bottom "privex", and press VOTE.

![](https://i.imgur.com/DFoBK4o.png)

Thanks to everyone for all of your support. Without the Steem community, we would have never managed to get so far.

If you'd like to grab a server, just go to our website: https://www.privex.io - we accept STEEM, SBD, LTC, and BTC as payment. We offer both Virtual and Dedicated servers (all the way from 1gb up to 256gb RAM, with 512gb coming soon).