Opt For The Data Express Lane With A LAN-Free Approach

Topics in this article

Have you ever played the game SimCity?  

Once you get slightly beyond beginner stage and build up a city, the biggest challenge becomes traffic.  

It becomes very frustrating, very quickly with all of these little simulated cars, until you realize the solution.

LAN traffic in your data center can be a real challenge too. But, there is an answer here as well and it can be found in the not readily obvious and not entirely new.

The SimCity car glut is much like the challenges from unbridled virtual machine sprawl and data growth.

You need a solution for getting data around that takes you off the highways and byways and onto the road less traveled.

SimCity Fantasy

In our fantasy world of SimCity, the most common solution is to build another, bigger road. But, the fix is temporary at best and only diverts the traffic. A better solution is to go with a train, but this is another above-ground endeavor that takes up precious real estate.

The best solution is to go in a completely different direction. Go underground with a smart implementation of a subway line, and get the train benefits with only a portion of the above-ground real estate requirement.

Local Area Network Reality

Back in the real world, the local area network is now so consumed by daily activities and data traffic that you simply need another solution instead of more networks. 

Do the words “LAN-free” mean anything to you?

If not, then you might not be in IT. Because if you are an IT professional, you know that most technologies rely on the local area network or LAN to some extent. 

You’ve got file shares, printers, wireless connections, unified communications, email, SharePoint, forums, Internet access, client/server communications, and so much more.  

The word “LAN” mixed with the magic word “free” often sends IT people into an almost euphoric state.

Sure, there is progress being made. Like a highway getting wider and wider, the network protocol limits will keep expanding from 1GbE to 10GbE to 40GbE to 100GbE and beyond. But today, before replacing your network infrastructure or waiting for the right host card, the easiest option may be to put the traffic on a different path altogether—the SAN path. 

Better Solution for Back Office Data Movement

Unless you have a dedicated network pathway for IT, your LAN has enough going on. 

Upon a thorough examination of your storage workload, you may discover that you use only 20% of your available storage area network (SAN) bandwidth at any given point in time.  

Now, I’m not suggesting that your storage arrays are sitting idle.

But, what this might allow you to do is leverage SAN-based copy techniques to push whole portions of data or replicate them bit by bit.

SAN-based data movement provides you with many potential benefits today with the promise of more tomorrow, including:

  • Operating system independence. Moving from the LAN to the SAN removes host dependencies such as operating systems.
  • LAN and server offload (no CPU overhead). Shifting the workload off of the LAN frees up valuable server CPU cycles for business processes, and for easier scaling of mission-critical applications.
  • Heterogeneous server support. Operating system independence begets server independence. Get the full benefit of commodity servers with no dependencies.
  • Rapid data transfer. Harness the untapped SAN pipe to move data more quickly from one point to the other. Remember the idle bandwidth cited above? Think analogous to rapid transit subway or tube.
  • Incremental data movement. SAN technologies incorporate differential technologies to copy only changed data. Think data de-duplication.
  • Application integration. It’s offered in some current technologies such as in the replica management provided by EMC Replication Manager.
  • Virtualization technology integration. Storage-oriented replication solutions can have integration points into virtualization technologies such as VMware.
  • Bandwidth reduction technology. Moving into the SAN opens up possibilities of leveraging storage-array based replication technologies and their follow-ons that use data compression and other techniques.
  • Adjustable recovery-point (RPO) and recovery-time objectives (RTO). Storage-based replication technologies can also allow for flexible point-in-time recovery should your data movement needs also encompass your disaster recovery requirements.

The Road Less Traveled

Using a SAN-based copying approach tends to work well when you need a quick copy of a production application, database or file system. It just makes sense. Instead of putting the request on the network, you offload the network by copying data within the storage array. 

However, using a SAN-based copying approach for larger data movement needs can be enlightening.

The time differences to complete network-based versus array-based copies are staggering. It can be several hours or days difference depending on the speed of the LAN and SAN links. 

I’ve seen a customer go from 16 hours to 30 minutes to make a copy of a large Oracle database.  Similarly, I’ve observed Exchange backup windows that were progressively increasing from 4 to 6 to 8 hour windows get slashed to 20 minutes—all done while the servers stay online.  

Now, unlike our SimCity solution, this approach isn’t entirely underground. It’s been around for some time and can be found in solutions like EMC TimeFinder and EMC SnapView.

The difference now is the role it plays in addressing the challenges of data growth and how it needs to evolve to meet the needs of today’s virtual environments and cloud deployments. Unlike SAN-based data copy technologies of the past, products that move data over SAN need to be able to prioritize delivery based on service levels (think gold, silver, and bronze), and at an application level and not just data.

Solutions are on the horizon and as you might expect from the SAN orientation, not necessarily coming from your traditional network provider. Data growth challenges hit all levels of the compute, network, and storage stack.

Look for SAN-based data movement up and down the stack, with integration into your virtual server layer, applications, and storage management.

About the Author: Mark Prahl

Topics in this article