FOOD FOR THOUGHT Multi-Site Help

[Sciens]

I am planning on a multi-site deployment. Physically diverse with VPN connectivity for WAN interconnectivity between sites.

I have a background in Cisco Voice and am wondering what solutions are available for an equivalent solution to Cisco Survivable Remote Site Telephony (SRST).

In my familiar environment, you have a central call manager cluster for server redundancy and if the WAN fails the SCCP (and SIP as well i believe) phones will fallback to a local Call Manager Express enabled Cisco router that usually has a PRI or similar physical line for local or emergency calls.

The only deployment I have fathomed to devise without SRST using iPBX is to deploy individual SIP registrar servers at each site and configure trunks to a pair of main site servers for call routing or enable a dynamic call routing function (I dont recall the exact name of this feature or technology).

Has anyone here come up with a solution or encountered a similar requirement before?

 

[tbrummell]

I approach this with 2 questions. Is budget a concern? How important is cost vs uptime/calling?

 

[tbrummell]

I see it as:

Option 1:

• Local iPBX 2021 (virtualized or bare metal, your choice) at each site
• if PRI is a requirement, T1 gateway on each site (I like AudioCodes but you may find something less expensive)
• if SIP trunking is allowed, SIP account setup with a provider for each site (assuming decent Internet access at site)
• IAX trunks set up between other iPBX hosts across VPN links
• a thought out dialplan that will allow inter-site dialing without the need for a prefix
• outbound telco dialplan showing company main number (or extension DID if you have a block of those)

Pro's:

• Maximum uptime per site

Con's:

• Increased administration
• More complicated dialplan
• More maintenance & hardware costs
• MAC's (Moves, Adds, Changes) are painful

 

[tbrummell]

Option 2:

• Single cloud based iPBX on a reputable provider
• PRI could be used in this scenario, would just take some finagling
• SIP trunk if allowed (and easier to accomplish)

Pro's:

• No IAX trunks
• No extra hardware
• No dialplan complexaties
• Less administration
• No hardware costs
• Minimal MAC's as users can just move their phone if they are switching locations

Con's:

• Site Internet outage drops phone service
• Cloud provider outage drops company phone service

There are many pro's and con's to both options. You need to evaluate what works for your company and how much money is being allocated. I've serviced both types of setups for major PBX vendors in the past, and it will always come back and bite you in the ass no matter which option you pick. Pick option 1, someone will complain about costs. Pick option 2 and someone will complain that their phones went down when they lost internet. So, damned if you do, damned if you don't. Pick your poison and make sure your presentation spells out the risk and benefit of each.

 

[kenn10]

A third option is two cloud instances in different cities/data centers and phones that allow for a primary and backup address for the server. Unless pricey PRI's are required, you can use a SIP provider, such as SkyeTel or BulkVS, which allows multiple routes for incoming calls in case of primary server failure. Outbound calling is not an issue with any of the VOIP providers if you set it up correctly.

I usually found that when our Cisco cluster went down or the link to it was lost, the phones would reboot and it took quite a while to come back in service on the backup system. With Asterisk and phones that provide a primary and secondary server to be programmed into them, fail over is more seamless than on Cisco. You still suffer dropped calls in progress but the phones switch to the backup location without rebooting.

It is vital that you have a broadband provider that has dependable service with adequate bandwidth for the number of calls at each location. You'll find that having a primary/backup cloud location is more efficient than a cluster of CM servers.

Now if high redundancy is extremely important and cost is no option, Asterisk servers can be set up more or less exactly like the Cisco system on site or across various cloud instances. You can have a cluster of call servers and a cluster of registration servers and a cluster of database servers. That is frequently done by large campuses and universities to handle thousands of extensions. This matches the Cisco approach to a large environment.

The required level of redundancy will dictate whether it is acceptable to have a site go down in the event of the loss of broadband link to a particular location. In my case, if the data link to the site went down, we used cell phones for emergency and we could not function function as a business anyway without internet access. The central cluster would still operate and take voicemail for calls that could not reach the isolated site.

 

[Sciens]

Well you guys are incredibly helpful!

All of the above confirms more or less what I was assuming. Some kind of PBX at each site is basically the necessity.

The deployment scenario is multiple store sites with their own existing POTS lines. So, I am led to believe that to meet my needs I could deploy a small form factor server (Intel NUC or similar?) for a local instance of iPBX paired with an ATA for outbound calls to the POTS line and intra-enterprise calls via IAX trunks.

I am not running call-center volume. Essentially adding a feature for internal 4-digit dialing between sites, voicemail and conference bridging. Likely rather low demand, so the above should meet me needs more or less.

Any input or modifications to note?

 

[Sciens]

I will also add, cost is really up to me as I am a consultant for this company and am looking for a good solution for their curiosity on how to improve as a multi-site enterprise.

 

[MGD4me]

Depending on how many sites you are tying together, you can use the 4-digit dial plan to your advantage. My scheme uses the first of the 4 digits as 'the office number', except that I don't start at '1XXX', since it becomes a bit more effort dealing with outgoing LD numbers.

Second, it is handy to have some common numbers in each office, For example, 2000, 3000, 4000, etc are all the main reception number. 2100, 3100, could be 'Sales', 2200 for 'Support', etc.

Just my $.02 ...

 

[kenn10]

Yes. I recommend standardizing your dial plan so each site knows the number of another site. Sears used to do this in their stores. For example, furniture departments in all stores were extension 201. Hardware was 209, etc. This encourages stores to direct dial other stores as required using your intrastore trunks as opposed to tying up outside lines or resorting to long distance calls.

If your stores are small (say 10 or less extensions), it is easy to concoct a standardized 4-digit dial plan where the first 2 digits are the store identifier and the last two digits are department or area of that store.

 

[tbrummell]

Yum! Brands did this with their Pizza Huts as well. Your Office USA was also the same, every office was identical in numbering.