New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
★ VirMach ★ Black Friday & Cyber Week 2018 ★ RAID 10 SSD ★ OpenVZ & KVM ★ Check inside for offers!
This discussion has been closed.
Comments
This black friday was busier, and with less manpower, sure. And the people creating tickets for what we already answered in knowledgebase or elsewhere definitely have to wait much longer since we do not have outsourced support.
About 3% of people are still waiting for their specials because it's running into issues.
We also still have weird bugs we're dealing with related to WHMCS in one way or another and that's definitely made things harder.
With that said though I do not think we need to change our strategy.
You need to hire more people just for 1 month, some part-time supports that can handle some repetitive tasks during this peak season could smoothen the whole process by a lot.
We already have addon for dedicated CPU, but I think with these new builds, essentially 2x the CPU power and 2x less people (2x less RAM and disk) we should basically be able to give "dedicated*" CPU out.
*As long as every single person isn't always maxing it out, we'll be fine.
Auto-throttle CPU is in the works but definitely won't be something that goes with it during launch.
Training will take longer than just doing it as it is. We did try hiring some people, but it did not work out after a trial period. We still have others lined up and people still applying. We just have time constraints there when it comes to training.
We're trying at the moment to filter out all the bad tickets and then make sure we're getting to customers that actually need urgent assistance first, and we're making good progress.
We even finally developed an auto IP changer.
Not necessary. The part-time supports are usually L1 and they can filter out most of the bad tickets so only the important ones are on L2 or L3. In that case you don't need to take 1 week or more to reply to tickets.
One of the most important thing in customer support is to let the customers know their progress, some of the options include queueing (like let them know how many tickets are in front of them etc).
If it is hard to give them an exact progress, at least some illusion of progress has to be done. L1 part-time supports can be best utilised in this case to let them know something is happening.
Expecting to use NAT64 ?
Sorry, I just didn't like the whole illusion thing and outsourced support didn't get much else done. We have a troubleshooter now and many more knowledgebase articles. We'll probably swoop in with some kind of automated bot that ends up being better than outsourced support but I don't think we'll be bringing outsourced support back.
Behind the scenes they also drained hours of our time daily.
Holy shit! Is that an easter egg I just found?!!!!!!!!
goes to underground bunker to check hypothesis
Thanks, I definitely woulda missed that one!
I see that mobo has built-in IPMI. Would you ever consider going with consumer grade hardware, with relaxed SLAs? Would it allow a lower price point or higher specs for same $$? (Thinking game servers, compute nodes, GPU or AI acceleration.)
Will you adopt N+1 clustering with scalable shared storage? Or do you prefer local-only storage for speed/simplicity?
Will there be a VirMach cloud?
Thanks for sharing, can't wait!
It will be meowing all the way.
Anyway, I'll take it with me.
Update: yup just found an easter egg, two actually! I reckon there's more released!
any hint?
I'll second FAT32's suggestion here, probably not through the exact same method, but being somewhat transparent regarding the progress...
For example, if sharing the number of "open" tickets (tickets waiting support input) is damaging or something, share the number of closed ones per day. I realize you're not always dealing with sane people, but normally, any sane person would be more understanding (as to why their 5 day old ticket is not being handled) if you're processing, say 300+ tickets daily, and taking care of the in them order they come.
Also, I don't know what percentage of your employees are in "holiday" mode, but people should expect a significant drop in productivity in any US company for +/-10 days around Christmas... I don't know if that's the case for you guys, but I'm guessing not many of your Asian customers would be setting their expectations accordingly...
I found 3 but all outdated
This is already kind of consumer-grade-ish.
We would have gone with Epyc if it's not basically double the price for half the performance, just because it's "Enterprise." Also sTRX4 definitely has its benefits such as higher memory support, but this takes the TDP way up so higher power cost and then concerns with the thermals. We might have had to therefore go with 2U, so the benefits of the double RAM basically cancels out.
There is another motherboard we tested out that didn't have IPMI, the ASUS Pro WS X570-ACE, and the benefits were that it supports PCIe 4.0, it has built-in 10GE, and it has RGB lighting. But the downside was that we had to get IPMI added to it somehow on top of the bulky NVMe RAID controller that has to sit there and block everything else. It would have also ended up costing $400 (or nearly double) price and we also had to find a separate chassis that would comfortably fit it with the PCIe riser in 1U. (Although, it actually is on sale right now, basically 50% off because it's gotten heaps of bad reviews.) Maybe we'll buy some more and mess around with it at this pricepoint.
GPU nodes, we've considered it. But it'll most likely come third or fourth in our colo product offerings. However, with virtualization I don't think consumer grade would work best there either.
For the NVMe SSDs, we are going consumer-grade, technically. The only real good enterprise NVMe SSDs are either outdated (several years old, slower) and the newest ones are extremely expensive. We're talking something like double price on what ends up being the most expensive set of components for the server.
What we have planned, and this is no way set in stone, is the following:
We're definitely going to try to add more of a "cloud" type backup/restore feature later. We have the logistics sort of worked out, but it's hard doing all this with SolusVM. It'd be a lot nicer if SolusVM actually did anything, but they won't since they're going to obviously try to push the expensive Solus.IO alternative.
We may start another company that offers actual cloud services with cloud-type features. Maybe using Fleio or something similar and ditch WHMCS/SolusVM completely for that.
(Edit) Oh and we plan on having basically one empty server ready to go per cabinet so we don't have situations where we have to rely on datacenter hands to do something major or situations that CC ends up being in a lot where they have no parts on hand.
We're trying to be as efficient as possible. We'd have to dedicate a significant amount of time to all this stuff and I would usually agree with you that it's better, but in this case it's just best we try to power through it.
We had something similar to this with an estimated response time and where it would kick in a message when it detected we're "busy" but as all things go, this caused a lot of negative attitude. We would get customers that would basically be pissed off before they even made the ticket.
We're understaffed. But this doesn't mean what you think it means, it just means I work 18 hour days basically.
After we got rid of outsourced, we lost a few other people because of various different situations and with COVID it wasn't easy finding replacements. We trained 2 new recruits recently but one didn't work out and the other got caught up with something else. There's a sys admin we have lined up to start work but then Black Friday happened and we're catching up to that first before we allocate time to training people.
4 outdated
Found 2 that's live but isn't what I need right now. I still can't figure out a third one.
My gut feel tells me that there's a big ticket easter egg in the wild now!
Huh, I only found the one.
Lucky you.
I even don't know the hunt game started
Another rotten Easter egg. Total 5. I have to sleep otherwise my brain will go bad too.
Wait, what? You found 5???
Yeah. But outdated
where is the snow?
It will show again when @VirMach appear