• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Data Center migration without any single downtime

S

sabarna_deb

Junior Member
Hi All,
We are in business of cloud computing. We are planning to migrate few customers from one datacenter to another datacenter without affecting any customer, no single downtime, no packet drop.
We are not handling any applications in servers, only infrastructure (IT). Kindly guide me how to proceed further? Kindly share your experiance? Specially i want to know from storage (data) perspective because i am basically from storage and backup. How we can transfer the data without effecting any clients and layout the same data into the original structure? We are having VMware, Clarrion etc. How we can advantage of VMware here........

Thanks
Sabarna Deb
 
Umm...

I once migrated a large number of clients from one data center to another such as you are proposing.

Let me know if your company needs an experienced consultant to help you with this project.

The answer to what you are asking is not a one liner.

Unless you want to hear "it depends".

Comblues
 
Actually it does depend.

Current technology does allow us the capability to extend a data center or entire network across Internet boundaries.

True, the architecture must be in place at first to support such a project or desired outcome.

Example: I once worked on a network that was in two different states. We used IBM for the SAN and had a point to point Ethernet hand-off from DC1 to DC2.

The global mirror updated across the these physical network hand-offs.

We only suffered a 24-28ms RTT and so we were able to transfer packets pretty fast over the 1000 or so miles between the two DC's.


More to come - there are technologies that support just this using VMWare's virtualization products.

Also VPLS is a reality and why not use OTP?

Think about clustering technology and other high availability technologies...

Very low latency and while we may lose a single packet we may not have to lose much more.

Comblues
 
If you use vmware on your servers, depending on how both your data centers are setup in your vsphere, this should be fairly easy for you to accomplish.
 
@comblues

What I meant was you cannot cut over to a data center in that regard with zero downtime and/or zero dropped packets.
 
Genx87 said:
If you are in the business of cloud computing. Shouldnt your staff know how to do this?
Click to expand...

Yeah... first post, asking for advice on a project that would in theory have millions of dollars in revenue hanging in the balance, with a company that would in theory have a couple dozen full-time engineers providing support, and many millions invested in hardware and software. I call shens one way or another.

I mean, I like anandtech forums and all that, but this is more a forum for "how do i get my gaming pings lower" or "should I get a ccna?". Not "how do I achieve a SP-class data center migration without downtime".

...and as a fully employed network engineer, THANK GOD that these kinds of problems can't be solved in a message thread.
 
Sure they are. Just click "vmotion all servers to standby cluster in another city" button.

Happens in seconds.

Now where's my 100k bucks for successful implementation?

OP - what you're talking about is possible, but takes significant bandwidth and complex infrastructure to implement that must already be in place. And it is significant, that's a severe understatement. I'm talking about another data center that is fully replicating everything from another right down to real-time memory, let alone storage and all the global server load balancers.

Depending on size, you'll need 4-10 gigs of bandwidth between them.
 
Last edited:
Comblues said:
Also VPLS is a reality and why not use OTP?
Comblues
Click to expand...

I think you meant to say "OTV".

We've done something similar, moving stuff from one DC to another w/ little to no impact.
We were lucky because we had dark fiber between the two DC's, and bandwidth wasn't an issue (multiple 10G links via DWDM).

Like others have said, the underlying infrastructure has to exist, along w/ staff & careful planning.
If you're just looking for a general directions & tips we can share our experiences.
But you need to hire someone to actually pull it off.
 
Railgun said:
@comblues

What I meant was you cannot cut over to a data center in that regard with zero downtime and/or zero dropped packets.
Click to expand...

No worries I think I understood what you meant. I was just trying to reference some items where the technology may allow for minimal downtime and packet loss. Some aspects of networking are there and others are still subject to the user's experience and circumstances.

I've lost a packet (one pip of a ping for example) just flipping from BGP to EIGRP or vice versa. One way I lose a pip and the other way I did not, for example.

Either way I did not lose my voice calls, which was the desired outcome of the process during a service loss.

So zero loss is a buzzword to me, but sometimes we can still lose a packet depending on the technology in question. My experience may be more limited than others.

Comblues
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top