VEEAM is ready for Enterprise environments ?

[liviu.tutuianu]

Dear All,

I've been using VEEAM for the last months. During this time, we implemented it in two countries and in the next period will be implemented in more countries. Because every country has different infrastructure and different requirements, we had to design different solutions.

Let me explain you a solution that was applied in one of our countries. So, we have 30 branches and one Head Office (HO). In each branch we have 2 ESXs and on one of them we have a VBR configured. This VBR manages only the replication jobs from each store. In each store we have 3 replication jobs. In HO, we have also a VBR installed that has a Data Domain storage attached. On this HO VBR we have a backup job from each store to HO. These jobs run every night and the data that is transfered is around 300 MB/ store.

Now, let me explain you the problems. Regarding the restore request, we have much operational work. Restoring a full VM for this country will occur maximum 1 time per year, but restoring files requests come weekly. Restoring a file from a replication job, need to connect to each VBR and start the restore wizard. After that, we need to save the files locally on the VBR and from there to copy the files to their original location. Also, if we need to restore a database that has files on more that one partition, we need to save files from each partition on different folders located on VBR and from there to copy them to their original location. In some cases, all these clicks take longer than the restore itself.

Also, take in consideration the fact that there are 30 branches. But, we have a country where we implemented in 120 branches. By the end of the year we expect to have VEEAM implemented in over 350 branches. This means that we will have over 1500 jobs. And by the end of 2013 we expect to have VEEAM implemented in over 500 branches! Well, this means a complex infrastructure that needs to be handled.

Also, for every country we need to grant access to certain people to monitor only the jobs from their countries. Using VEM doesn't allow us this.

So, here come my 3 questions:

1. Is there anyway by which the files/folders restoring process could be optimized? Maybe a central console from which to make a restore very fast when it will be required? Also, the restores to be made to a certain destination, not only on the VBR. Because we need to restore files especially from Replication jobs, 1-Click Restore feature doesn't help us too much.
2. How can we define different access levels for persons from different countries to monitor the status only for the jobs from their countries?
3. Because backups contain also very sensitive legal data, is there any way by which the persons that will make the restore will have only the restore access and not reading or modifying the restore data?

Hopping that some of you got until the end of this threat, I am looking forward to hearing from you some ideas.

Thank you in advance,
Liviu

[Gostev]

Hi Liviu,

Both points 1 and 3 are addressed with our 1-Click restore functionality. From what you are explaining, I believe you are very unique at what you are doing (replication only). Thing is, replication is NOT designed to be a complete replacement for backup, because it does not provide sufficient retention and uses too much disk space (which is especially important for environments of your size). We have designed our replication is designed to provide very low RTO and a handful of restore points for selected VMs, but keeping in mind backup should be performed for all VMs (including the ones you are replicating) to cost-efficient protection with extended retention replication will not be able to meet, as well as great amount of restore options (when replication is designed primarily of full VM failover).

Thus, I would recommend that you consider performing backups to achieve requirements 1 and 3 - this would probably be the best, since you need the solution right now. And in future, we can consider adding 1-Click FLR to replicas as well - if this becomes a popular request, which I do not see right now.

Regarding point 2, to achieve separation, we recommend deploying separate B&R "domains" for where such separation is required - in your case, one B&R install per country, for example. I am assuming we are talking about may be dozen of countries, right? So, this would be good immediate solution, and it has been working well for other customers. However, as always, we are looking for feedback on what works well and what does not, so we could further improve our product (without turning it into uber complex monster, of course).

Hope this helps!

[liviu.tutuianu]

Hello Gostev,

First of all, thank you for your quick reply.

Well, using replication in every branch is very OK because the retention period for these jobs is only of 7 days. The vital data are being backed-up from branches to HO and are configured in 3 set types: daily, monthly and yearly. And the data from HO is being replicated to an offsite location. If we will use only backup jobs, then we will get nothing better than the solution that we have now. Besides I can say that it will use more disk space and will take more time to complete, as we don't have the option to select certain files or folders to be backed up. So, I cannot say to the client that I offer him a more expensive solution, that does the same things like the old one (meaning only backing-up data) and making their job harder (as it takes more time to restore files from each replicated VM).

So, I offer him a more expensive solution, but he also gets with this one a DR solution (having the possibility to start anytime a full working replica in each branch) and backing-up their critical data in HO. From there it's being replicated to a secondary site. Also for replicating successfully their database servers, I had to develop certain scripts that integrated into their existing systems and satisfied their business needs. Now I am at the point to offer them a restoring and monitoring solution at least better than the existing one.

Now, consider that I am your customer that buys you hundreds of licenses and you need to accomplish my needs described above. How would you proceed?

This is why I was really wondering if VEEAM is really ready for Enterprise Environments. If it doesn't do all the things like the actual enterprise solutions + doesn't bring new features, then it's quite hard to say that is very to Enterprises. Also, I still have doubts regarding the long term data archiving. We have on country in which they want to keep data for 10 years (as this is their legal requirements for invoices...). And for sure, we don't want to use tapes!

Thank you in advance,
Liviu

[Gostev]

Hi Liviu,

I am sorry, but I cannot really add anything to my previous response. With the current version, goals 1 and 3 can only be achieved when doing backups (not replicas). This may change in future if we get many similar requests, but I realize you need solution now, so I provide you the best possible resolution that is available *today* - instead of trying to sell you futures.

Also, I think you are confusing "if VEEAM is really ready for Enterprise environments" with "if VEEAM is really ready for MY environment". As I've said, what you are doing (replicas without backups) is very unique, which is why you have run into this problem in the first place.

Generally speaking, Veeam got its first 1'000+ socket customer over 2 years ago, and since then we've kept adding more and more enterprise features with every release (and more and more enterprise customers) - and will not stop doing that. So yes, our product is definitely ready for at least certain Enterprise deployments.

To be honest, we do lacks a few major features of typical "enterprise ready" product:
- We are not complex (do not need for 3 months to deploy, and an army of PHDs to maintain)
- We are not expensive (reasonably priced and no need to pay for each feature)
- We are not stuck in old vision of data protection (we solve the same old issues in different - simple and elegant - ways)
I don't think we will ever add those features though

Thank you!

[andersonts]

Liviu it does indeed seem that you have some fairly unique requirements. Most replication solutions are not built with file recovery in mind...at least not the ones I see in customer environments I support that range from 2 sockets to thousands of sockets. You certainly can get this functionality if you spend enough of your customers money which is clearly not your goal here. That being said you touched on a couple of points that might help us define an even better strategy for you. For instance you mention only needing data for 7 days and only transferring 300 MB per night. If that is the case then why not run a backup job before or after the replication job completes? Since our backup jobs dedupe and compress the data you will not need a ton of disk to perform this...the 300MB per night you are transferring for the replicas will be much less than that with compression and deduplication. That will give you the best of both worlds without costing your customer anything more than some extra disk space. Another option would be using backups instead of replicas and if/when they needed to spin up a replica they could use instant restore which would give them only slightly longer RTO and use much less disk space. This is certainly not even close to the same as other backup solutions out there which still require much longer to do a full VM restore and cannot even offer replication as an option. Just a thought on that...perhaps not the perfect solution but very good options.

I agree that products can always improve and I think our dev folks certainly do a great job of staying ahead of our competition. The thing I like about working at Veeam is the features we do have are generally light years ahead of our competition and they work well as opposed to having a bunch of features that have tons of caveats and "if you do that then you can't do this".

Some thoughts around the 10 year data retention requirement that your customer has. I agree going to tape is the least feasible option...at the very least there is no guarantee that the tapes will be readable in 10 years! The idea here would be that your customer needs an archiving platform. If you imagine that typical data usage patterns are that 85% of data is never touched again after the first 45 days of creation then how about using an archiving platform to push that data off to cheap spinning disk? Customers may get concerned about needing policies and different departments need this that or the other retention cycle but really all you need to start is a policy moving everything after so many days...you can figure out when to age it out later. For the 45 days that the data is on tier one storage you would want to use the best tool for the job (Veeam in the virtual world) to back that data up. Once it is on archive storage the archive software can take care of it from there to keep it sorted out, safe, and replicated to a secondary location. Imagine how happy your customer would be if you save them 80% of their tier one storage budget and use that to implement an archiving solution, cheap archive storage, and Veeam as the ultimate DR tool for their virtual environment. If you insist on using your backup tool as an archiving tool you can still achieve this with Veeam you just need to work with us to develop a strategy that works for you and your customer. Customers are already saving lots of $$ today writing Veeam compressed and deduped backup files to tape...since most legacy solutions cannot write deduplicated data directly to tape this is a huge advantage for Veeam. Imagine 50 tapes in a weekly that is kept for a year going down to 15 tapes a week = 35 x 52 x $50 per tape = $91,000 in annual tape savings alone!