Pretty cool -- slightly larger than my home lab

  • Thread starter Deleted member 138126
  • Start date

Deleted member 138126

D

Deleted member 138126

GLiCE7N.png

I love seeing this sort of thing. 3.34THz of CPU, 20.62TB of Memory...
 

Deleted member 138126

D

Deleted member 138126

OP
Hehehe Enterprise Plus for the lot. Consists of several clusters (the screenshot is from the point of view of the Datacenter), including a large Citrix farm. Just finished upgrading everything to 6.5. 10G everywhere.
 
Associate
Joined
9 Aug 2004
Posts
2,061
Location
Sea of Dirac
yeah so far ie and u1 have been the most stable for me. don't have chrome but suspect it works fine too. glad I held off 6.5 till u1 was out as so far it's paying nicer with the dell and hp gear I have.
 
Associate
Joined
18 Oct 2002
Posts
1,765
Location
Kent, UK
I just spent the last couple of days standing next to this bad boy:

TD1.jpg

TD2.jpg

TD4.jpg


The only moving parts are the fans, although they did make a bit of a racket :D . A lot of enterprise-class SSDs in there (12TB each I think) making up the storage capacity.

Not quite as 'big' in absolute terms as the OP's data centre VM capacity, however to cram this much database performance into a single cabinet is :eek:

Of course, this single cabinet is pretty small (in storage capacity terms at least) when compared to many Teradata customer installations (which run to many PB) but it definitely represents 'bleeding-edge' in terms of database performance - and of course can scale linearly as far you like by adding more nodes/cabinets.
 
Associate
Joined
18 Oct 2002
Posts
1,765
Location
Kent, UK
Wow I bet that runs Oracle queries REALLY fast. Very nice.

Haha - nothing runs Oracle queries really fast (well, unless they are really simple and someone has indexed and optimised the *** out of the underlying tables). Teradata hardware only runs the Teradata database software, which is in a different league of database performance for complex multi-join analytical queries.
 
Soldato
Joined
26 Nov 2002
Posts
6,852
Location
Romford
Nice stats.

The consolidation ratio on these clusters doesn't seem great. 2534 VMs over 413 hosts, that's just over 6 servers per host - why so low?
 

Deleted member 138126

D

Deleted member 138126

OP
Nice stats.

The consolidation ratio on these clusters doesn't seem great. 2534 VMs over 413 hosts, that's just over 6 servers per host - why so low?
Mine was empty because I took the screenshot before migrating stuff across from the old 5.5 cluster. It's now sitting at 992 VMs on 70 hosts which is 14:1 -- and there are still a few more VMs that will migrate over, so it will end up around 15/16:1. It means we have plenty of overhead so don't have to stress about capacity management (can afford to review it on a quarterly basis).
 
Soldato
Joined
26 Nov 2002
Posts
6,852
Location
Romford
Some of our hosts have 40+ guests, but we average about 30:1. Hosts run at about 80% memory capacity without any sharing and the CPUs are generally low. You must have some beefy VMs or are you running VSAN so needs all those nodes for storage?
 

Deleted member 138126

D

Deleted member 138126

OP
I'd say we average 60% memory utilisation. CPU is always pretty low. Running at 80% RAM is uncomfortable for me, you're getting close to a situation where you lose a couple of hosts and all of a sudden you're close to the 90s, especially on the smaller clusters).
 
Back
Top Bottom