Hello community,

As you may know, not too long ago, we reached the Top 20 Witness spot and are happy to have had the opportunity to consolidate our position in this ranking. Below you can read about our Witness activities across Hive, Dlux, SPK Claim Chain, and more.

=============================

Our Hive Witness: @threespeak

Privex HIAB (Hive Node in a Box),
16GB RAM, Intel Xeon CPU

===========================

DLux Witness: @speak-lux

Ryzen 7 3700X
64GB DDR4 RAM

============================

DLux SCC Node: @speak-scc

Ryzen 7 3700X
64GB DDR4 RAM

============================

HIVE API Node

Ryzen 9 5950X
128 GB DDR4 RAM

We also provide a public HIVE API node for the community. Available at https://hive-api.3speak.tv/

Storage Infrastructure Operation

Starting with storage, for the previous 3.5 years, we have maintained native storage servers along with our legacy storage system. For the native storage, that has grown to around ~100TB for our ever-growing IPFS storage needs. We add approximately 1.5TB of data per month in new videos across ipfs-cluster and the legacy data storage system.

We currently pay in the thousands of dollars per month for 3Speak's storage infrastructure. As the SPK Network evolves to support Proof of Access, CDN daemons, and more, this will help bring down our operational costs at a core level. We are also looking at other forms of optimizations to decrease operational costs.

Moving to Community Lead Peer to Peer Video Storage

As the SPK Network advances and Proof of Access storage rewards are released, the community will have more tools to self-host and operate their own native video infrastructure. As self-hosting becomes widespread, 3Speak will be looking to reduce the number of videos stored on its own servers and rely on self-hosted infrastructure.

We plan on doing a slow phase-in period of shifting to reliance on self-hosting, while we test/improve stability and other critical pain points. We will make a formal announcement before making any major changes to 3Speak.tv's backend storage that might directly affect users. When proof of access is available, we will look at backing every video on 3Speak with a minimum amount of BROCA incentive to help migrate over.

The Proof of Access and Peer to Peer encoding systems are expanding. They are currently in a beta test phase. We are making many improvements and fixing bugs. There are now 27 separate encoder nodes operating at the time of writing this.

IPFS Garbage Collection Optimizations

Present day IPFS can be very slow due to numerous pain points. One of those pain points is garbage collection. For most people that simply means removing unnecessary data from the machine. But, for us, it plays a critical role in the operation of 3Speak and SPK Network. For example if we are unpinning some content or doing a migration we need the ability to quickly reclaim space. We accomplish this using some minor architecture and code changes on the go-ipfs binary. We will have a blog post out next week that will go much more into detail on the changes and provide some basic benchmark comparisons.

What changed:

• Pre-index all IPFS blocks into a database creating a giant graph database
• Significantly reduced the amount of data that needs to be read from disk
• Fast in memory indexes
• Benchmark results will be published once available. Anecdotally and theoretically it should be faster than default IPFS.