Am 27.08.2013 00:07, schrieb Jorge Bastos:
>>> Hum, which disk's setup do you have on the HOST?
>>> HW RAID ? RAID, 1? 5?
>>
>> http://h10010.www1.hp.com/wwpc/il/en/sm/WF06b/12169-304616-241493-
>> 241493-241493-3971510-3957473.html?dnr=1
> 
> So you're storage is an iSCSI, right?
> No latency?

no latency, the controllers are Athlon 2600+ 1600 Mhz, have
plenty of write-back cache and not to forget that VMware
is damned good at caching at the layer below the guests

two LUN's

*  1.6 TB RAID6 for public servers (8x300 GB SAS 15000 rpm)
*  5.5 TB RAID5 for internal infrastructure (4x2 TB SAS 7000 rpm)

at normal operations each LUN is one one of the two controllers
due controller-firmware-updates or failover both are on the same
which happens transparent and only vCenter whines about degraded
redundancy, as long there is nothing degraded not to forget
that *both* storage paths are used and so you have practically
2 Gbit and spreaded load

2 dedicated SAN switches, two DL380 (G7/G8) and any network connections
are redundant and load-balancing - the setup exists in this form since
2010 while one host from 2008 was replace din the meantime

well, in fact you have for sure virtualization overhead but on the
other hand you have much more powerful hardware as you could pay
for 30 physical servers and most of the time a lot of guests are idle

not to forget fully redundancy and de-deuplication backups on local
disks on one host which are freezing the guest-FS at the begin of
a backup and make a snapshot of the guest - at the moment of freeze
you may have some latency and lose a few network packets, but that's all
_____________________________________________________

http://www.vmware.com/products/vsphere/features-data-protection

current de-duplication 2.33 TB to 272 GB (fileserver-storage excluded
and daily offsite-backup per rsync, the same for any other data
additional to the DataRecovery inhouse)
_____________________________________________________

we started with VMware in 2008 (one host, local storage), dbmail
in production started 2009 and 2010 a second host, vcneter-Server
and the storage and after 5 years i would never ever go back to
physical machines

*never* compare Virtualbox or VMware Workstation with a bare
metal hypervisor - on my workstations and homeserver i run
VMware Workstation and for normal operations it may be eual fast
but if it comes ot load no way to compare, hence i had a Load
of 136 on one of the guests last year while the others did
run well and even the attacke done was very responsible in
a ssh-sessiion, my phsyical machines are practically unuseable
with a load above 10

Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
DBmail mailing list
[email protected]
http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail

Reply via email to