Comprehensive data protection for all workloads
Post Reply
sureshkumarbb
Lurker
Posts: 2
Liked: never
Joined: Jul 28, 2019 5:49 pm
Full Name: Suresh
Contact:

Understanding the scope of Veeam with exagrid

Post by sureshkumarbb »

Hello Techies,

My client was experiencing higher RTO/RPO with Netbackup (DD as storage) and so looking for suggestions from our data protection team.
Is there any way I can achieve this with Veeam or Exagrid or with both in combination for below questions? I am a kind of complete newbie to Veeam and exagrid and please apologize me if I asked a complete irrelevant question here. While going through youtube videos, I find some hope with Veeam which I want some more clarity from you experts. Thanks for reading this.

Question 1)
How to protect and recover the Oracle Database of 50TB. The RTO for the recovery is 30 mins and RPO is not more than 5 mins.

50 TB Oracle DB last successful incremental backup at Sunday 9 Pm. The DB corrupted on Monday at 03:00 PM in the afternoon.

Is there any way I can achieve this with Veeam or Exagrid or with both in combination?

Question 2)
Can we pretect a VM which is of some 100 TB with RTO as 15 mins and RPO not more than 5 mins?

Question 3)
There is a Fileserver of 1 PB. How efficiently we can backup this? 8 hrs is backup time and 20 TB recovery in 30 mins? Is it possible?

Question 4)
Does Veeam has any user frindly interface that users can themselves recover their lost files without the help of admins?

Question 5)
cost effective method to store 500 TB front end data (monthly backup) with 7 yrs retention. Does exagrid is best solution here to me here?
HannesK
Product Manager
Posts: 14288
Liked: 2877 times
Joined: Sep 01, 2014 11:46 am
Full Name: Hannes Kasparick
Location: Austria
Contact:

Re: Understanding the scope of Veeam with exagrid

Post by HannesK »

Hello,
and welcome to the forums.

First of all: I suggest that you contact your local Veeam rep and involve the local SE team in the design as there are many components which are relevant. I just expect now that your network, storage, CPU, RAM and everything is "right sized".

1) RPO of min is possible with log shipping every 5min. To restore 50 TB in 30min you need about 230 GBit/s restore speed. I doubt that this is possible with classic restore mechanisms (no matter which kind of backup storage device). If it is a VM, you might want to look at Instant VM recovery but even with that, I fear that the log replay would take longer than 30mins

2) RTO with 15min should be possible with Instant VM recovery. RPO of 5min is harder. If the infrastructure is fast enough to do backup every 5min... I mean it takes already some minutes just for the VMware snapshots... I don't see 5min RPO as realistic value.

3) Windows? Is it a physical machine? How many disks (I ask because it needs parallel disk processing)? Any filesystem larger than 64TB (VSS requirement)? Change rate (how much data needs to be processed in 8hrs?) 20 TB in 30min would require 91GBit/s on average, well, let's say 100 GBit/s with some overhead... I would call it not realistic.

4) It depends what kind of users you mean? For end users, there is no Veeam interface (usually customers use "previous versions" which is built in to Windows). For VM / physical machine admins, there is an easy to use web interface or you could use the agent itself on physical machines

5) Please understand that we cannot advice on licenses in the forum. But in general, Veeam does not care about the amount of data from a licensing perspective. The main question is, whether you do a full backup every month or whether you just do incremental forever for 7 years. If you go with incremental forever, then deduplication appliances are usually not the most cost effective way (no matter which vendor). If you go with monthly active full, then a deduplication device (no matter which vendor) can be cost effective. With 500 TB it might be cheaper to go with two stages: "normal" backup repository + object storage (cheaper)

Best regards,
Hannes
sureshkumarbb
Lurker
Posts: 2
Liked: never
Joined: Jul 28, 2019 5:49 pm
Full Name: Suresh
Contact:

Re: Understanding the scope of Veeam with exagrid

Post by sureshkumarbb »

2) RTO with 15min should be possible with Instant VM recovery. RPO of 5min is harder. If the infrastructure is fast enough to do backup every 5min... I mean it takes already some minutes just for the VMware snapshots... I don't see 5min RPO as realistic value.
>> I understood that snapshot processing takes time. But will Exagrid takeover the changes in 5 mins once snapshot is ready?

3) Windows? Is it a physical machine? How many disks (I ask because it needs parallel disk processing)? Any filesystem larger than 64TB (VSS requirement)? Change rate (how much data needs to be processed in 8hrs?) 20 TB in 30min would require 91GBit/s on average, well, let's say 100 GBit/s with some overhead... I would call it not realistic.
>> Assume we have a physical machine with 1 petabyte and a VM with 1 petabyte. I mean lets consider 2 scenarios for possibility here. And both have parallel disk processings with many disks (as 1 petabyte will be on many disks). No filesystem larger than 64 TB. Change rate is zero as we are intending to take backups in night hours where there will be zero and if not very low change rate to the existing volumes.

5) Please understand that we cannot advice on licenses in the forum. But in general, Veeam does not care about the amount of data from a licensing perspective. The main question is, whether you do a full backup every month or whether you just do incremental forever for 7 years. If you go with incremental forever, then deduplication appliances are usually not the most cost effective way (no matter which vendor). If you go with monthly active full, then a deduplication device (no matter which vendor) can be cost effective. With 500 TB it might be cheaper to go with two stages: "normal" backup repository + object storage (cheaper)
>> Can you please explain what exactly is normal backup repository here. Is this a data that got deduped and compressed?
What exactly is object storage you referred to here please? (if its cloud, then it has to be ignored as we are not intending to go to cloud storage for compliance reasons).

Thanks for the initial reply and for your time in answering this. These are actually just the initial requirements posed by my client and anything like right sizing of CPU and others are well managed by others. I am here for data protection keeping the pain of other things to other teams. So, I am drafting now the possibility and scope different backup vendors which I find Veeam is good to be considered.
HannesK
Product Manager
Posts: 14288
Liked: 2877 times
Joined: Sep 01, 2014 11:46 am
Full Name: Hannes Kasparick
Location: Austria
Contact:

Re: Understanding the scope of Veeam with exagrid

Post by HannesK »

2) If I estimate 100 GByte change rate per backup, and 0,1h (well, that's 6min), then the backup speed is about 2.3 GBit/s. Estimating a data reduction of 50% it will be around 1.2GBit/s single stream for that VM to exagrid. Exagrid might be able to answer the question which model can deal with that, but I expect that they can do that.

3) The change rate is between two backup runs. So it does not matter whether there are many users logged in at the time of backup. I estimate 1% change rate per day. That means 10 TB data to backup in 8hrs. Backup speed on average would be 2.8 GBit/s. That sounds like "no problem" from Veeam perspective. Of course, the initial backup would take longer. All incremental backups are block based "incremental forever" backups.

3a VM) direct SAN mode / backup from storage snapshot - go for it :-)
3b physical machine) should also be possible today. But keep in mind that parallel disk processing for the agent is planned for the next version (not available today)

5) A "normal" repository is a "bunch of disks". Usually multiple high density servers / storages with 60 or more 7.2k disks on 4RU with REFS. REFS is important for performance.

I always calculate built-in dedupe / compression with 50%. Exception: the customer knows that he has tons of un-compressable data (videos / pictures / encrypted data).

With object storage I mean S3 compatible storage systems. Customers say that they are cheaper (depends on amount of data and vendor) than normal storage systems and allow 3-way mirror. These systems are getting more and more popular, which is why I suggested it to think about that.
ChrisSnell
Technology Partner
Posts: 126
Liked: 18 times
Joined: Feb 28, 2011 5:20 pm
Full Name: Chris Snell
Contact:

Re: Understanding the scope of Veeam with exagrid

Post by ChrisSnell »

Hello Suresh - Veeam and ExaGrid are designed to give you the best performance possible from Veeam and a dedupe storage target. If you DM me with a few details I'd be more than happy to put you in contact with someone local at ExaGrid who can work with you in detail on these questions. Dedupe backup storage will certainly be more cost effective than using fast disk, and I'd have great reservations on something like ReFS being able to scale to those numbers.
Post Reply

Who is online

Users browsing this forum: Semrush [Bot] and 273 guests