VMware Cloud Community
idiriasl
Contributor
Contributor
Jump to solution

Remote working centres and HA?

Hi all,

I have this situation:

- A client has three different work centres. One of them (lets call it A) currently stores ERP databases and server and the others (B and C) work on it through TCP/IP connections.

- The centre A sometimes suffer from electricity downtimes so centres B and C are not able to work.

- The managers want people at any centre to be able to work on the data even if one or two of the other centres fail.

I had though on mixing Vmware and HA with NAS replication. One NAS at each centre so if A fails, B can take control and continue working against B NAS. Nevertheless, I can not figure how to do this or where to start and even if HA is the most apropriated approach... perhaps DRS?, Do I need any specific storage hardware for this?.

Clarification: They do not need zero down-time, some minutes are ok.

Clarification 2: What about Storage Appliance?

Thanks in advance.

0 Kudos
1 Solution

Accepted Solutions
a_p_
Leadership
Leadership
Jump to solution

... centre A sometimes suffer from electricity downtimes ...

To clarify. Are these planned or unplanned downtimes? How long do these downtimes last? Would an appropriate UPS be an option?

If the downtime is unplanned, there's always a risk of data loss or corruption and regarding the vSphere features you mentioned, it's HA which does initiate a restart of a crashed VM. However, to be able to restart the VM, the shared storage needs to be available. So you need a storage solution which provides synchronization (synchronously if possible). Most major storage providers offer such solution. With most solutions you need to activate the mirror though, before it can be accessed, so that may not work with HA. There are also other solotions like Lefthand (now HP P4000) which provide a transparaent failover.

Depending on your needs, SLAs and budget, you need to decide what will be the solution that works best for you.

André

View solution in original post

0 Kudos
9 Replies
AndreTheGiant
Immortal
Immortal
Jump to solution

Welcome to the community.

About VSA it is limited to max 3 nodes. And they must be in the same network (are all the site on the same network?).

With VMware HA you need a shared datastore (you can have some trouble with replication).

Have you considered instead of use Veeam Backup & Replication (or similar product) to replicate the VMs across the sites?

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
idiriasl
Contributor
Contributor
Jump to solution

Thanks Andrew,

I don’t fully understand how would such application help me… I mean, imagine the VM1 is running on site A and I have such an application replicating the VM… Is this a “real-time” process?, what state will have the machine once restarted on site B?, how does this connect to VMware Vsphere (How do I tell VSphere to provide HA under such scenary?)… Too many questions perhaps?, I am totally lost L.

Regards,

Jose.

0 Kudos
a_p_
Leadership
Leadership
Jump to solution

... centre A sometimes suffer from electricity downtimes ...

To clarify. Are these planned or unplanned downtimes? How long do these downtimes last? Would an appropriate UPS be an option?

If the downtime is unplanned, there's always a risk of data loss or corruption and regarding the vSphere features you mentioned, it's HA which does initiate a restart of a crashed VM. However, to be able to restart the VM, the shared storage needs to be available. So you need a storage solution which provides synchronization (synchronously if possible). Most major storage providers offer such solution. With most solutions you need to activate the mirror though, before it can be accessed, so that may not work with HA. There are also other solotions like Lefthand (now HP P4000) which provide a transparaent failover.

Depending on your needs, SLAs and budget, you need to decide what will be the solution that works best for you.

André

0 Kudos
AndreTheGiant
Immortal
Immortal
Jump to solution

I don’t fully understand how would such application help me… I mean, imagine the VM1 is running on site A and I have such an application replicating the VM… Is this a “real-time” process?, what state will have the machine once restarted on site B?, how does this connect to VMware Vsphere (How do I tell VSphere to provide HA under such scenary?)… Too many questions perhaps?, I am totally lost L.

If you use Veeam, Vizioncore, or similar solutions the replication is scheduled (usually you can get around 15 min of delay). But can be VSS consistent.

To have "real-time" replication you must go to storage replication.

Andre

Andrew | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
idiriasl
Contributor
Contributor
Jump to solution

Thanks,

Those are unplanned downtimes. I had thought about implementing a UPS (many people suggests that :-))... but it seems everybody wants virtualization nowadays... I will have a look at the mirroring technologies and involved costs... The problem is that connection between centers is done by wifi networks :-(.

Regards,

Jose.

0 Kudos
idiriasl
Contributor
Contributor
Jump to solution

Andre, just another question. When you mention HP 4000, do you mean the hardware or virtual one?. I suppose both of them do the same, but if you have any cooments on it...

Regards,

Jose.

0 Kudos
a_p_
Leadership
Leadership
Jump to solution

Actually I was thinking of the physical nodes. Whether the virtual VSA could eventually be an option, depends on the workload, IOPS etc. Btw. if a virtual VSA is an option for you, don't forget to take a look at VMware's new VSA which was released with vSphere 5.

André

0 Kudos
idiriasl
Contributor
Contributor
Jump to solution

Thanks André,

I am afraid that my question remains open... The system is not very loaded by a lot of users but I still don't know how does the HP P4000 does for "real-time" mirroring and what impact it would have on networks (remember links are made by wifi connections). I suppose P4000 does the job at a block level which would be the only option for me but it is still a mistery what impact would have to run 3 P4000 SANs on such an infraestructure... will have to go deeper on this, perhaps contacting HP.

Regards,

Jose

0 Kudos
a_p_
Leadership
Leadership
Jump to solution

Yes, I think talking with the vendor or a partner/reseller would be a good way to find out about the possibilities and the requirements. It certainly is important to know the requirements (e.g. latency) to see whether you need to probably redesign (other parts of) the infrastructure to achieve you goal.

André

0 Kudos