Radoslaw,

I think you don't understand RPO correctly.

read:
https://www.enterprisestorageforum.com/storage-management/rpo-and-rto-understanding-the-differences.html

Tony Thigpen

R.S. wrote on 1/31/20 11:24 AM:
W dniu 31.01.2020 o 07:07, Timothy Sipples pisze:
Radoslaw Skorupka wrote:
To complement and clarify: when using physical tapes (*
see below) your RPO and RTO may be 36 hours or zero.
No, your RPO certainly won't be zero. A backup is a (hopefully useful)
representation of data as it existed historically, at some particular past

moment(s) in time. It takes some amount of time to run a backup -- let's
call that "minutes or longer" for working purposes. Backups run at
periodic intervals -- let's call that "hourly or less often" for working
purposes. Your backups, without something else, facilitate a best case RPO

that's as long/big as the maximum (worst case) time elapsed since the
start of your last good backup. That practically always(*) means a RPO of
"a couple hours or more."

A long/big RPO usually holds RTO back too, but there are a few rare
exceptions. On the other hand, it's quite possible to have a long/big RTO
with a RPO of zero.

(*) Why not "always"? Exotic, contrived exceptions might be possible, such

as custom software that synchronizes writes directly to local and remote
tape.

No, my RPO *is* zero. We do not talk about backup itself, which is ...the same for VTS, MAGSTAR, USB stick, etc. Despite of backup media the backup is representation of historical data. What I'm talking about is time delta between primary and secondary (local, remote) backup copies. And again: some (typical?) modes of VTS replication are not synchronous, so there is time delta between local copy is closed (and marked as done) and remote copy is also finished. For duplex write in HSM both tapes are being written simultaneously, so end of backup job means you have both copies ready.

Note: Earlier you mentioned that PTAM makes RPO longer. Yes, it *adds* the time to "historical representation". VTS non-synchronous mode adds the time also (much less). but HSM duplex write add zero.


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to