Grids' distribution

[ivan98]

Hi I'm looking for guidance on deploying of Applogic grid(s) on multiple data centers.

Should I've one grid over multiple data centers (over a single virtual L2 network), or should each data center has its own grid? What is the deciding criteria? What should I consider?

For discussion purposes, let's start with 3 data centers: A-B is 30ms, B-C is 102ms and C-A is 131ms. BW is 100Mbit/s all round. If insufficient, this can be increased with practically zero cost. Offering-wise, I'd offer to my customers such form of "off-shoring" in case of disasters.

[Jsmart]

Hello,

In today's environment most likely even with good bandwidth this would not work for a single grid. the latency is a little too much for the back network. I have seen this tried even with 10GE networks and not work correctly.

Generally what you would build would be independant grids in each datacenter and you build failover and redundancy into the applications on these grids that run your "service" to give you not only failover but true fault tollerence.

If you would like to go into more details on what you plan to offer I could probably make some more meaningful suggestions.

Jessie

[arisythila]

yeah. I've tried this.

We had 10GB Dedicated line with great providers going from Seattle, to Dallas. Latency was just too high to get this to work correctly. Even if 100Mbit/s is received it was still too high.

I tried this with lit fiber. you maybe able to do this with Dark fiber tho. (you light both ends)