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.
Comments
What kind of redundancy are you interested to achieve?
You can have a range of options, including storage replication. Each step up to 99.999% make it complex and pricey
Curious if there are any smaller ones that do clustering IE A&B data is written to the active and backup, and if there is any problem with A, then B takes over without a hiccup.
I heard theres no point in them.
I had a VPS with LeaseWeb and they used Sans, barely managed to get over 20/MBs
I have experience with fiber channel SAN (SUN/Storagetek 6xxx and Hitachi AMS 2100/2300/2500) and ata over ethernet Coraid.
Synchronus replication is a license feature for FC and require some middle appliance with coraid.
Both option hurt your wallet
I use both the old Storagetek and the Coraid SRX3200. Both have full hardware redundancy but I don't use any replication feature. Costs will be very high even for midrange vps.
For some DIY storage with replication I use DRBD but I think that with the I/O you have with lowend high density vps performances will suffer.
In general SAN are not the speediest thing for the much loved dd tests, but they sustain low latency even with high load and this is good for applications.
In our server room we use Dell MD3200i (iSCSI) and Fujitsu Eternus DX80 (FC). The Dell MD3200 is a LSI logic design, also sold by IBM. Both have dual controllers in active/passive configuration. The second controller automatically goes live if the first fail. This actually happened on one of the MD3200i and worked perfectly. I also used the dual-controller feature on the Fujitsu SAN to perform a rolling live firmware upgrade. (The Eternus stock firmware had a very nasty bug, and caused downtime). I also set-up a scheduled replica from the primary SAN to the backup SAN using Veeam. As Prometeus said, the next step is a syncronous replica between storages, but it is a bloody expensive feature. I was surprised by the low latency of both SAN's under load.
Msa 1000, msa2000, eva1440.
As prmeteus said, msa isnt really built for speed, its built for stabiliy.
you will get aprix sale speed from it with many async read/writes but very rarely see any higher values even if your the only one on it
Yes I find ioping a better measure of VPS performance than writing a bunch of data to the disks.