Orhan Ergun 4 Comments

Active-active data center design can be accomplished in many ways.

You want to have best possible data center design, don’t you  ?

I will mention from some technologies which are used in an active-active data center design at the end of this article.

Businesses want to have continuous operation even under failure condition.

You might build a datacenter with redundant devices, links, power sources.

But if entire datacenter fails for some reason ( Fiber cut,  big power outage, disaster ), your operation stops.

Or if you are operating some applications in a distributed fashion, big part of your operation stops.

If operation stops you loose reputation.

reputation

Active-active data center design requires at least two data centers, most of the time very close to each other. (Synchronous replication and clustering has very low latency requirements )

Active-active data center design requires extra budget, skilled engineering,uncommon and sometimes immature technologies ( I will explain them in another article related on this topic)  thus different design thoughts.

First active-active data center design model :

If your company decides to build or colocate a datacenter or even get a service from public cloud provider to implement its private datacenter in a cloud.

Rather than running all applications in one datacenter with another as backup, you can distribute and run the applications on more than one datacenter. 

In this case, you run some applications in one data center, and some in the second.

Second data center will not be sitting as idle thus waiting the primary to fail.

This operational model is called active-standby, used in a context of disaster recovery.

Second active-active data center design model:

Same application can operate from both datacenter in a scale-out fashion.

This operational model is implemented with the help of load balancing, stretched clustering, LAN extension, state synchronization on the stateful devices ( Firewalls, IPs, Load Balancers and so on ), IP localization and Mobility technologies( RHI, DNS, LISP).

It is quite complex but provides excellent RPO and RTO.

Both are considered an active-active data center model because the two datacenter are actively used thus resources are not waiting as idle.

Data loss and recovery time are expected to be zero or near to zero.(RPO and RTO).

The design of an active-active data center model is different from a compute, storage, networking , applications and network services standpoint than active-standby data center.

It’s Your Turn

How many data center your business has ?

Do you have a disaster recovery or disaster avoidance strategy ?

Let’s talk about your strategy in the comment box below.

 
0.00 avg. rating (0% score) - 0 votes
  • Roy Lexmond

    Nice post, we designed an external network with more specific routing to preffered DC.
    We can switch DC’s by inserting more specific routes if required else it goes automatic and loadbalancers behind the routed frontend manage vips and pools with servers in any backend DC.

    Design alternative we had in mind was RHI but after comparing pros and cons for this specific network we went for more specific routing.

    • probably you would like to talk about this design 🙂

  • Pingback: HSRP VRRP GLBP Comparison()

  • Pingback: BGP Best External | Cisco Network Design and Architecture | CCDE Bootcamp | orhanergun.net()