More than 2 nodes

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

Post Reply
petr
Posts: 13
Joined: Sun Feb 27, 2011 2:55 pm

Tue Aug 09, 2011 8:53 am

Do you plan (and when) to offer more than to 2 node HA? For example 4-5 nodes which will offer more speed and availability?
User avatar
anton (staff)
Site Admin
Posts: 4021
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Tue Aug 09, 2011 10:23 am

Yes, multi-node config (sync and async) is coming. Soon. Very soon :)

It does not work as you expect BTW. With more then three nodes doing synchronization with each other to keep distributed caches coherent the whole thing saturates network and brings extra latency killing performance. So it's pretty unexpected to see more then 3-4 nodes in homogeneous config.
What's going to work is using multiple redundant nodes keeping shared cache as a gateway to second layer of nodes acting as a storage rather then processor/cache nodes. Take a look at Fujitsu grid storage systems. They are built with referenced architecture in mind.
petr wrote:Do you plan (and when) to offer more than to 2 node HA? For example 4-5 nodes which will offer more speed and availability?
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
petr
Posts: 13
Joined: Sun Feb 27, 2011 2:55 pm

Tue Aug 09, 2011 10:30 am

Hi,

that is what I'm asking for. Our farm is not as fast as it should be with one node down, so another backup or MPIO node will be very needed. Do you have ETA? How it will work?

Thank you
User avatar
anton (staff)
Site Admin
Posts: 4021
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Tue Aug 09, 2011 10:38 am

We'll provide configuration with no performance degradation on partially broken cluster. ETA is this Fall. You may get early experimental builds sooner.
petr wrote:Hi,

that is what I'm asking for. Our farm is not as fast as it should be with one node down, so another backup or MPIO node will be very needed. Do you have ETA? How it will work?

Thank you
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Post Reply