Good news, there is a way and it’s called the Cloud.  I recently did a proposal for a client that is using hosted exchange for 75 users and growing.  They are looking to bring the exchange in house by using Exchange 2010.  To do this they are going to need:

1.  New servers to accommodate Exchange 2010
2. Storage to accommodate at least 10gb per mailbox
3. Rack space, power and cooling to be upgraded
4. In house expertise to administer Exchange
5. In house expertise to support end users of Exchange
6. Proactive maintenance of new hardware
7. Procure Microsoft Exchange and Windows licensing

In this particular case there is just a single in house IT Supervisor to handle the new Exchange, along with all of the other aspects of the network, spanning multiple locations.  This is a perfect play for a public cloud infrastructure.  We can eliminate all of the above needs by configuring a solution that brings both the hosting and the support over to the cloud.

How do we do that?  It’s simple – we create a Cloud vOFFICE that includes a front end vSERVER, vFIREWALL, vINFRASTRUCTURE, and backend vSERVER for exchange.  This is a fully functioning environment with managed DNS Hosting, 10 Mpbs internet, etc.

Simple diagram of what this infrastructure looks like (click on it to enlarge)

There is no hardware to buy, there is no software to buy, there are no maintenance renewals to worry about, there is no depreciating of hardware over 3/5 years, there is no long term contract, no end users to support (on exchange in this scenario), etc-  I could go on for a while but I think you get the point.  Now the conversation has changed from a capital expenditure to an operational one.  The cloud is an operational expense, not a capital expense, and possibly can come out of a different budget and save the capital expenditure budget for other in house initiatives.

The cloud is a very powerful tool to rethink how we deploy applications and hardware.  We need to embrace it, implement it, and grow with it.

Marty Gallagher

marty@ampdatasystems.com

Before recently I would completely dismiss the idea of talking about SRM if the customer had 2 or 3 Vmware servers.  I felt, “no way are they going to go for SRM for the price in such a small install.”  What a stupid thing to say.  Instead of explaining the benefits of SRM with DR/BC, I let an opportunity to add value go and with it a customer that could have had a lot of their problems solved with one product. 

The benefits of SRM are clear and with all the focus on business continuity and disaster recovery these days, it’s a perfect fit to complement those solutions.  With SRM you can not only create DR plan you can actually TEST it.  Let me say that again, you can TEST it.  How powerful is that?  Many companies have DR plans in place most likely in the form of a word doc put in a binder and stored at multiple locations.  With SRM, you have a complete virtual machine DR plan that can be tested once a month or once a quarter or whenever you want.  The best part is certain manufacturers allow you to do the testing while replication of the VM’s are taking place from the production site to the recovery site.

SRM integrates with SRA adapters that are specific to certain manufacturers.  It does not move the data but rather leverages replication technologies through SRA adapters.  SRA adapter integration is available from a wide variety of manufacturers.  Be certain to look for manufacturers that can failback operations to the production site after it is back online.  This is a key differentiator when deciding what manufacturer to go with for storage for the VM’s in your environment.

Yes SRM will add to your capital expenditures(keep in mind you only need to license the production site if not doing bi-directional replication), but it will greatly reduce your operational expenses and pay for itself after you have done your first SRM TEST recovery.

Gartner put out a report that said by 2013, 60% of data center servers will be virtualized.  You need a plan to handle that server sprawl.  SRM is the perfect tool for the job.

AMP has designed SRM solutions for clients with all different types of storage.  Even if you have heterogeneous storage, AMP can still design a fully supported SRM environment.   A great way to lower your CAPex and leverage what you already have in place.

Next week I will talk about how to protect physical servers and put a DR test plan in place to protect  those physical servers.

Marty Gallagher

Marty@ampdatasystems.com

The question I ask all the time is “if you had unlimited budget, what would fix first in your data center?”  The answer I hear the most – “our backups.”  A close friend and much respected colleague in the industry recently made the comment that doing backups are like taking out the trash.  It has to be done, but nobody wants to do it.  That is right on about how most people feel! 

 If clients could afford it and had that unlimited budget, I would design a backup/recovery solution that encompasses CDP, VTL, staged disk, DeDupe, and tape.  All are part of an enterprise backup/recovery solution depending on your applications and your companies RPO/RTO.  If some of the applications need an RPO of 24 hours, then you can get away with tape (there are other options but to keep this simple if you backup once a day your RPO is 24hours because you are willing to lose 24 hours worth of data in case of a disaster).  If other applications require almost zer0 RPO then go with CDP.  What it really comes down to is tiering your applications and then setting backup/recovery polices for each of the different tiers. 

There are certain products that incorporate a mix of backup/recovery technologies on the same device.   VTL is often combined with DeDupe to offer high speed multiplexing to virtual tape drives and then simultaneously remove common blocks and store to disk.  Now with OST becoming the norm for backup applications to interact with these appliances, it makes the master/media server aware of where the data is when it’s backed up.  Very powerful technology!

Here is a simple way you could look at what backup/recovery solution would fit the best:

Tier 0 – mission critical, zero RPO, and minutes for RTO

Solution – CDP – go back to a specific i/o’s and be up and running in seconds/minutes

Tier 1 – critical, RPO/RTO hours, and need data available onsite because of compliance/retention policy

Solution – CDP withVTL/ Dedupe – snapshots allow clients to pick a point in time to restore back too and then can dedupe the CDP data for onsite availability

Tier 2 – business process data and file system data that isn’t critical

Solution – stage to disk then offloaded to tape – very easy to back up to staged disk and eliminates your tape backup window over the network

Tier 3 – non-mission critical application data that don’t need low RPO/RTO

Solution – tape – possibly only needed to backup once a week or once a month. 

Keep in mind that there are ways of combining the Tier’s into one complete overall solution.  It may look like a lot of management but there are some highly efficient products that can eliminate backup windows and make restoring simple.

Backups are like taking out the trash, but at least there are ways to make taking it out sexy!

Relationships first, business second.

Marty Gallagher – marty@ampdatasystems.com

How many times in the last couple of years have you been on a call with a customer, mfg, or partner and heard the term deduplication when talking about backup/recovery/replication?  Heck, we have heard it so many times that we actually turned a word into a commodity.  Everybody throws it out there like it is and will continue to be the greatest technology ever.  I have to respectfully disagree.   I will give dedupe 3 props –

1. It’s a cool technology
2. It has a fantastic sales pitch behind it
3. It fits customers who need onsite retention in a small footprint and bandwith reduction

 It has become such a common term that it has become like the “thin provisioning” buzzword of years ago.  EVERYBODY has it, so it is just expected to be a feature.  This is where dedupe is now!

 With dedupe integration on popular backup applications at the host or media server level it seems as if some of the luster of appliance dedupe has been lost or is in the process of being lost.  Couple that with big name storage mfg’s having dedupe built into their platforms, it could mean that dedup will be just another checkbox that customers want to see – and you know what?  There is nothing wrong with that!

The next wave of backup will focused on recovery solutions rather than backup solutions (and you will start seeing products called Recovery Management Solutions and new term for CDP).  When you think about it we always should have asked not “how are you backing up your data” but “how are you going to recover your data in the event of a loss.”  It’s not about backup, It’s about RECOVERY! 

In 11 years of talking to clients, the backup has always been the bottom feeder of the datacenter.  There was never any time to do it, nobody really wanted to do it, and it was always the first thing cut when budget wasn’t there.  And this is no knock on my clients, they were pulled in multiple directions and over tasked and told to make do with what they had. (which we all know is the same slow, unreliable, tape backup device)

Be sure to check back as part 2 will examine how to make it about recovery (RTO/RPO) and minimize tape (as much as I would love to say eliminate, it still isn’t a reality just yet for most clients)

Enjoy the Masters and lets hope we don’t hear about Tiger and his transgressions as much as we have heard about dedupe in the last couple of years!

Lets AMP your infrastructure!

Marty Gallagher – marty@ampdatasystems.com