Whether someone uses our Node-in-a-box packages (NIAB) or not - doesn't tell you much about the witness's actual skills, and what they bring to the platform.

Our NIAB packages are popular with both "noobs", and a few veteran top 20 witnesses too.

The witnesses in the top 20 who use us, generally use their NIAB server as either a backup witness node, and/or a public seed node (though I believe there are 1 or 2 within the top 25 who use one as their primary witness).

Lower ranking witnesses (those below or at rank 30) generally use their NIAB server as their primary witness node, I believe mostly because of the low cost combined with convenience, which is great for witnesses who make less than $100-200/mo.

What really matters when choosing witnesses to vote for, is:

• Is the operator of the witness savvy enough to:

  • handle upgrades
  • fix their block_log / shared_memory when things go wrong
  • understands how to broadcast their witness_update transaction with at least two different tools (sometimes one tool might be broken after a HF, while the other works fine, so it's important to be able to use multiple tools incase their preferred tool breaks)
  • understands how to setup a pricefeed, inc. how to set a bias on their pricefeed during extended price disparaties with HBD

• Is the operator's witness reliable? They should:

  • Have at least ONE backup witness server ready at all times.

    • NOTE: If you decide you want/need more than 1 witness server, please make sure that each server is:

      • In a different country, or at least a different city/state than your other servers.

      • For extra safety, you may want to use a different server host too, though generally each region of a server host is mostly isolated from the other regions, so an outage
        affecting multiple countries of a given server host is rare.

        By using @privex - this point can be covered simply by buying servers in different regions, e.g.

        • our DE/FI dedicated servers are provided by a third-party datacenter/network,
        • our US regions are provided by three different third-party DCs/networks unrelated to the DE/FI provider
        • while our SE (and soon NL) dedi's + VPS's are 100% first-party (we own the hardware, network equipment / cables, and IP space).

      • Note that you should be mindful to avoid known "bad" hosts (e.g. OVH's network is too unreliable/unstable for Hive witnesses),
        as well as hosts which are just a reseller/subsidiary of the same provider you already use for your other server(s). (e.g. OVH, SoYouStart, and KimSufi are all part of OVH and use the same DCs + network).

    • For witnesses in the top 20, backup servers are pretty much essential if you want to stay in the top 20. I recommend having at least 3 witness servers (1 primary, 2 backups), or more if possible. I personally have 4 nodes (1x DE, 2x FIN, 1x SE).

    • For witnesses between rank 21-35, it's very important to have at least 1 backup server, due to the fact you'd have a high chance of becoming part of the top 20 with very little notice (and you won't stay there if your server fails within a few hours/days of being in the top 20, while having no backup to get back online ASAP).

    • For rank 36-50, a backup server is a good idea to have, but not too urgent. At rank 40, a witness would get ~23 blocks per day, or close to 1 block per hour. If you have a dedicated server with a fast single-threading CPU, then it may be possible to restore a native Hive snapshot in under an hour, before your next block.

    • For witnesses below rank 50, a backup server is not as important (due to the long gap between blocks) - as you'd have plenty of time to either repair your server / installation, or spin up a temporary backup server (@privex's NIAB servers are great for those) to keep the witness producing until their primary node is back up.

  • Have an automatic failover system, such as @therealwolf 's Hive Witness Essentials

  • Setup bots which alert you if you miss blocks, or your pricefeed hasn't updated in over 3 hrs, such as @deathwing 's Friday Discord Bot

  • If your witness is ranked 25 or higher (especially for the top 20), you may wish to employ a "backup admin" - another person who will monitor, maintain, and fix your Hive witness when things go wrong and
    you're asleep / on vacation / at work etc.

    A backup admin should be someone who's Linux savvy, understands how to manage, maintain, and fix Hive nodes; PLUS you trust them not to steal your account / coins, as they'd likely require your active key to manually
    switchover your signing key if needed (though this can be done with your current signing private key with some tools now), as well as to be able to re-enable your witness after temporarily disabling it to protect
    you from missing blocks.

    EXTRA TIP: You may wish to employ someone who's in a timezone at least 8 hrs ahead/behind your own, which may allow them to monitor your witness while you'd normally be asleep.

• (one of the most important points) Do you think that the person(s) behind the witness are making valuable contributions to Hive in some way? Bonus points if their witness node appears to be operated by someone who actually understands what they're doing.

This is a bit long for a comment, but I hope it helps people who aren't sure what qualities to look for in a witness, and at what rank certain redundancy/reliability measures should be expected.