Network differences in the same DC

gsrdgrdghd

I've recently gotten around to setup SmokePing to monitor my VPSs and noticed something odd:

You can see that the server with GetKVM has a nice smooth line and the one with Providerservice has a lot more jitter although they are both in the same DC. Obviously this doesn't have any practical impact as it's only +-5ms but i'm curious what could cause this.

(us2 monitors from BuyVM Buffalo, fr1 from PyramidServers (OVH?))

Jack

Maybe Providerservice oversells the port to hell?

gsrdgrdghd

@Jack said: Maybe Providerservice oversells the port to hell?

I don't think thats the case because the graphs were similar durign the night (when there was probably very little traffic)

GetKVM_Ash

This is strange, they are all in the same suite etc. I was also thinking that it could be the occupants on the node at PS causing it, but im not sure.

gsrdgrdghd

Well the node is probably quiet oversold, it's from their 2€/month 512mb XEN offer. Maybe its also due to the differences in virtualization technologies

GetKVM_Ash

@gsrdgrdghd said: Well the node is probably quiet oversold, it's from their 2€/month 512mb XEN offer. Maybe its also due to the differences in virtualization technologies

Yeah, i took one for myself for backups from our infrastructure (Canceled it yesterday), was not impressed. Constant server reboots with not much reason as to why in the announcements other than a "Misconfiguration".

gsrdgrdghd

Well i can't really complain about the node (09), 99.93% uptime so far. I think there were 3 reboots since i've gotten it but i'm not running anything important on it anyway (just gameservers).

GetKVM_Ash

@gsrdgrdghd said: Well i can't really complain about the node (09), 99.93% uptime so far. I think there were 3 reboots since i've gotten it but i'm not running anything important on it anyway (just gameservers).

The one i had was on 09 as well. I cant trust something that's up and down with our WHMCS/SolusVM DB backups :P

JoeMerit

Node 9 should be fine now, they were fixing a problem with the ipv6 configuration .

ErawanArifNugroho

I'm also on Node9, and having several emails about reboot. 8/16/2012 - Emergency reboot 9/6/2012 - Reboot because misconfiguration

Everyone may have problem, but if too frequently, 1 every month, maybe I will leave them

FRCorey

My traffic does not really drop at night

Mon5t3r

@gsrdgrdghd said: PyramidServers (OVH?))

yes. :)

interesting, i knew it since Ash start selling VPS on their location. Ash network seems more great than them. i have one server with providerservice, but i can't say much since it's a "fresh from the oven" server.

ProviderService

I have to admit that we had several initial problems with our Xen setup. Some were hardware related and some were software related (bugs in the Xen kernel and SolusVM for instance and also one configuration issue regarding IPv6 neighbor solicitation). While fixing the last known problem yesterday, one node crashed, alas. We are very sorry about the downtime caused.

As we are using really high end server hardware now (with redundant power, redundant air shrouds and of course redundant discs) and all software issues are supposed to be fixed, we do not expect any further problems.

As for the jitter issue gsrdgrdghd mentioned: Neither our network nor our hypervisors are oversold. Currently I do not have an idea where the differences are coming from but we will definitely have a look at it.

Mon5t3r

WOW!!!

About
Username ProviderService
Joined July 8
Visits 7
Last Active 8:52PM
Roles Member
Thanked 1

i smell LowEndExpansion here.. glad to know you're here @ProviderService :)

Jack

I know we have;

Soluslabs Licensepal Jay

All we need now is John or Matt from WHMCS.

William

Same DC can or cannot matter, the AS number is no indication how the traffic is routed as the AS number can be used at multiple locations with completely different upstreams (i.e. Softlayer Amsterdam and SL WDC)