If you have ever lost data because the snapshot was taken too far from the point you wanted to get back to, if you tried to bring up your DR site and found out application cannot start because of inconsistency, you probably want to check out DellEmc Recover Point.

Recover Point – what is it?

Recover Point is an appliance – physical or virtual that is taking care of your storage LUNs replications.

With Recover Point you can create local (on the same storage) or remote (on a different storage) copies of your LUNs.

With a unique mechanism the replicated LUN can be mounted to any point in time and even to the exact point of any block change.

What makes Recover Point so special?

I’m replicating my data to the DR already, what different with Recover Point?

• While in most of the replications software you are scheduling your replication, with Recover Point every block that is written to the LUN is immediately replicated to the target.
• Recover Point uses a unique mechanism called “journal” that lets you go back to every block change, not only to the points in time scheduled. (we will explain more about this mechanism further down).

I’m taking a snapshot every minute and I keep it for 3 hours, how can I make it better with Recover Point?

• By taking a snapshot every minute for every LUN you are using a lot of system resources which can impact your storage performance. When using Recover Point you don’t use storage resources.
• When using snapshots, if you get corruption on the LUNs, occasionally you can’t recover with snapshots because it depends on the source LUN. When using Recover Point you actually have a clone of the LUN with no dependency on the source LUN.
• The granularity of your snapshots is every minute. With Recover Point you can mount to any point in time

How does Recover Point actually work?

Let’s talk about RPA (Recover Point appliance) and Recover Point clusters:

On each source or target storage, we install a cluster with 2-8 RPAs. If we need a replication between storages we can connect the clusters. Each cluster can act as source, target or both. 

Recover Point uses the “splitter” that is installed on DellEmc storages. When activating a source LUN for replication, the splitter duplicates every block that is written to the LUN. It then writes it on the LUN and on the RPA. Only after both are written and the splitter get acknowledgement. does the storage send write acknowledgement back to the server. Once the RPA get the write request, it immediately sends it to the target.

So far it all seems nice but let’s find out what happen on the target side once the block is received on the target RPA:

The target for the replication is built from 2 components: LUN at the same size of the source for data and additional LUN called journal for the changes log. Once the RPA needs to write a new block on the LUN it first write the change on the journal. Only then does it write the new block.

By using this log file we get the ability to roll the target LUN to the exact point in time we need. The amount of time we can roll back to depends on the journal LUN size (can be any size) and the amount of writing to the source LUN.

Customer story

Before I will tell you about 2 special editions of Recover Point I want to share a short customer story showing that by using Recover Point we don’t lose any data in case of a crisis:

Well the story is about a customer using a virtual windows file server, all data written on RDM disk form a storage protected with Recover Point copy.

The story begins when the customer realized that a ransomware was running on one of the user’s computer, making a lot of files on the windows file server encrypted.

The first reaction on the site was panic, but then we started to work:

• The first thing we did was check the server’s audit logs and find the computer that was running the ransomware.
• After disconnecting the infected computer from the network, we tried finding out when the first file was encrypted.
• We then mounted Recover Point target LUN seconds before that specific time but still found encrypted files. We continued doing this, going back a few seconds at a time, until we found the point where the files were clean and unencrypted.
• When we had the exact point before the file encryption started, we rolled back the changes to the source LUN.

So at the end of the story the server was down for only around 30 minutes, no data was lost and the ransomware had almost no effect.  The customer was amazed at how simple, fast and effective this process was.

Recover Point for VM

Recover Point for VM is a special edition of Recover Point that instead of replicating LUNs, its replicating virtual machines.

RP4VM (Recover Point for VM) is storage agnostic because it works with VMware Vcenter and esxi hypervisor. RP4VM uses virtual RPAs and Instead of using storage splitter RP4VM install a splitter on each esxi on the cluster and the management software on the Vcenter as add-on.

You will get all capabilities as Recover Point have in the terms of local (same VMware cluster )and remote (different VMware cluster on the same or on other Vcenter) replications and any point in time granularity. You just have to make sure you got space on your datastore for both target and journal.

Recover Point Metro Point

Recover Point Metro Point is another special edition of Recover Point designed specially to use with Vplex Metro. (You can look at my blog about Vplex in order to get the idea about it).

Since Vplex Metro is using virtual LUNs mirrored on two different storages there is a question about how to replicate the data, local and remote.

It’s a little complex to understand but once you get it the implementation is quite simple.

• First step is connecting two Recover Point clusters – each of them connected to one Vplex arrays from the Vplex Metro. By doing so you can create a local copy on each of the storages connected to the Vplex.
• Second step, in order to get remote replication is to connect a 3^rd cluster on another site. The replication uses one of the Vplex cluster as active and the 2^nd as standby. In case of any failure of the active Vplex, the standby path will become active automatically.
• The complex scenario is when you have Vplex Metro both at the source site and on the target site and you want to replicate from Vplex Metro to Vplex Metro. (virtual stretched LUN to virtual Stretched LUN)
  • In this case you have to create the first step on both sides.
  • Then you to connect all 4 clusters, 2 on each side.
  • The replication will have one active path between the sites and all other paths will be on standby.

So as you can see, the granularity, accuracy and simple setup and management are all reasons we choose to use Recover Point above all other continuous data protection products.