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 ★ RYZEN ★ NVMe ★★ $8.88/YR- 384MB ★★ $21.85/YR- 2.5GB ★ Instant ★ Japan Pre-order ★ & More
This discussion has been closed.
Comments
We're scheduling maintenance to replace 3 failed disks in Tokyo, with new ones so we can generate the services. Emails will be sent out soon, these should essentially just be a quick reboot.
We're considering having a different line of service with 10Gbps line that we sell at a premium. I have noticed that when a bunch of tiny plans go on a server, no matter how good the network blend is or the hardware, it's just not able to keep up with the tons of what I assume to be VPN traffic. The packet levels end up being insanely high, similar to a tiny denial of service attack. The NIC just isn't built for that and I've tuned the settings for it to be relatively stable, but still.
As for hardware, the disks we're keeping are stable. The rest we're removing.
@Virmach Having a bit of a 'mare in Seattle too.
Could this happen to storage plans too?
If you're talking about SEAZ001, I finally ended up making them realize what's being requested after having a meltdown and they've been taking an entire day so far to fulfill an emergency maintenance request since.
Unfortunately, I can't have clones of myself at every facility, yet.
Unless people end up using storage nodes as seedboxes, probably not, and even then, still no, because they have a different highly functional NIC. Even if they use it as seedboxes and it ends up being the same less capable NIC, still no, because it'll be a lower quantity of total people.
This would be fantastic, I really miss my 10G burst from CC. Was a super nice little perk when I was moving backups off the box, etc. Would happily pay a bit more for that again.
Is this INAP's remote hands or does DediPath have staff there? INAP's hands have always been great for us and pretty knowledgable, but I suppose it varies per DC and we haven't been in any of their Seattle locations.
Yeah, we mostly skipped over "clustering" our nodes up at the beginning. I do have around four built 10Gbps nodes right now, I'm trying to decide where to put them. The options I've been considering:
I think it's a combination of DediPath trying to schedule things to save money, and the added "middle-man" effect that comes in with communication. We had the same problem when we were under one of CC's "subsidiaries" if you could call it that. Once we moved direct to CC, it improved.
The saving money part is fine when it doesn't absolutely drag out to an insane degree, especially for an "emergency" rate ticket.
If I don't get an update in the next few hours we're going to just assume they lost the disk or whatever... and proceed with re-creations for now.
I think this might be the best case scenario but I'd be cautious with offering unmetered, a few people usiing 1-2gbit constantly can degrade the network for everyone else. (Also check PM if you have the time, sent over some info about INAP, etc)
As mentioned some pages back, it's no big deal for me as I have a forlorn hope to migrate it to NYC.
I'm just a tad surprised at how long SEAZ001 is u/s.
I want to do something to help VirMach but have no idea
That makes sense but it's still not ideal, espeically for higher priority tickets. Might make sense if you're not paying for the hands (complimentary service, etc), but if you're paying then that's unacceptable. The communication makes sense but it's just a drive swap. I feel like that should be extremely trivial vs a board replacement, specialized equipmnt, etc. Maybe I'm just expecting too much though.
Well it'd be fair use and if they cause network degradation we'd talk to them about it. But I guess you're right, unmetered might still paint a false picture. How about this, if we go with the Chicago idea, we'd still advertise it as however many TB, we'd just sell a specific plan that doesn't get suspended immediately if you go over, and we waive overages. Could be interesting.
Send to NYC.
From what I captured from the VM ethernet, there are lots of pkt which are not supposed to be sent to/through my VM, those packets have neither destination nor source targeted to my machine, and furthermore, those are more likely a SYN attack from outside to your network blend, I think that may also impact the node networking performance?
Yes, I've noticed that as well. Not a lot that can be done for it but we're discussing some improvements that we can try doing. We do have a small layer of protection at the host node that we can try beefing up a little bit.
that explains why my ping at TYOC040 is fluctuating
I still haven't applied some of the NIC optimizations to that node yet. It's somewhere on my list.
Maybe a plan that's speed capped but unmetered? Like 250mbps unmetered, etc. Might be useful for people pushing large amounts of data without worrying about overages, etc.
and there's no update on this yet.
My vote is send one per main location, I'm with LiliLabs, the 10gbit burst was really nice to have. I'd personally rather have a fixed amount of transfer in my preferred location than unlimited fair use.
Seeing that virmach has launched a special event, I hope that I can also be lucky to win the prize, invoice! #1417908
Okay, we'll do it this way and send it to all locations that also have a storage node.
It's probably worthwhile to have every node connected at 10g on the LAN even if public network is limited to 1g, if you ever roll out internal networking the cross machine connectivity would be amazing to have, especially for locations with the storage nodes.
my Invoice #1402140
Node DALZ004 network seems to be working now, sweet!
Performance is awesome. Better than expected. Thanks VirMach!
Pumpkin Positive!