As far as I know, a 5340 Netbackup Appliance can handle 134 Deduplication concurrent backup streams.
In my environment, we are going to use 2 of those appliances. We have almost 80 SQL servers from I which we have to take transactional backups every 8 minutes (some cases every 5 minutes)
We haven't tested this yet because we just racked the appliances and we are about to install the master server.
Does anyone know o have any suggestions about this?
All the SQL Servers are VMware virtual servers running on a Nutanix Architecture.
We will use 10Gb LAN.
When you said you had transaction backups to run every 5 minutes, the first thing that popped into my head was, "So, what happens when you have to reboot the appliance ? "
Or what happens when it's time to upgrade and the appliance is unavailable for running jobs for 3-4 hours ?
I'll also point out that generally backup speed is limited on the Client-side, not the Media Server-side of things. If those database servers are under heavy load and you kick off a transaction backup you may not finish within 5 minutes & will end up running more than one job at a time (backing up different logs of course); is that okay with all involved?
Or my favorite example, if a pipe in a bathroom above the DC springs a leak and soaks the rack holding all of your backup gear (because of course the DC folks put all of your gear together, that's "your" rack), will the database servers survive until new hardware can be deployed ? Or will they all shutdown due to lack of space for those frequent transaction log dumps ? Is your answer still true if this happens on a local holiday with minimal (or no) staffing ? =)
What are you doing to get this critical data offsite, tape or replication ?
Unless you deploy NBU with an eye towards HA, it should not be considered a HA application in any way, shape, or form and neither should any of the backups it runs. I generally tend to tell DBAs that any of their deployments should be able to handle an 8 hour (or however long it would take me to recover that particular Master's catalog) outage.
Fun food for thought.
Thank you for your time.
About the downtime, we will have another appliance to handle the jobs.
Probably we will configure STUG for these jobs.
I´ve asked the customer to have a centralized location in order to put the "sql dumps" and then we can back up the "backup files" as files while the appliance is down.
about the critical data question, we have storage from hitachi to use as a private cloud in local DC and we have another 5340 appliance in remote DC for replications. In the remote Dc, we have a tape library as well