locked
Multi Site Deployment RRS feed

  • Question

  •  

    I am in the information gathering stage for a OCS deployment. My main hang up is how communication is handled between multiple sites. I follow how the edge servers are deployed and communicate with each other. I'm just not getting how internal user communicate between sites

     

    1.   In the microsoft guide it seems as if the there is an pool in each site. How do users in site A talk to users in SIte B.?

    1a.  In the multi site microsoft guide is it assumed there is a VPN link between the sites?

     

    2. When connecting remotely though the Access edge in site A then next hop is the director. If you are are user whos pool is in site B how does the director know which pool to direct you to?

     

    3. My company has a single domain, single forest. Since you can have only on SRV record per domain for automatic configuration. How would you configure DNS so users connect to their local pool.

     

    4. To alleviate these problems would it be better to just deploy one pool in a main site and have all other sites connect though a vpn site link to the main site. In this config how would you have users in say Site B & C user there own local AV Edge and Web COnferencing server.

     

    THanks for any help you guys can offer!!

    Thursday, March 6, 2008 3:00 AM

Answers

  • The answers to these questions generally encompass the discovery and architecture phase of an OCS deployment engagement, but I'll try to give you some concise answers.

     

    1.  Microsoft's documentation generally assumes that you have a private WAN link beween sites. This is commonly acheived by using things like MPLS connections, but VPN is another option (though there are increased risks that need to be accounted for).

     

    2.  The Director looks up the SIP URI in Active Directory, finds the pool associated with that user, and directs the traffic accordingly.  This is identical to how users communicate between sites - remember that the Director is just a regular OCS server that doesn't contain users.

     

    3.  This is a common discussion item.  You can architect that very basically by pointing everyone to a Director in your headquarters or be as complex as using location-based DNS referrals through something like an F5 solution.  The drawback in the first example is a single point of failure while the latter example is expensive and more complex.

     

    4.  OCS is usually centralized except in very large delpoyments.  This is not a problem for IM, but the A/V and Web Conferencing experience can be negatively impacted by this (especially A/V) if there is a large amount of latency or low bandiwdth between sites.  It is not possible to place A/V and Web Conferencing servers by themselves in a particular location - you must deploy an entire pool.

    Thursday, March 6, 2008 1:54 PM
    Moderator

All replies

  • The answers to these questions generally encompass the discovery and architecture phase of an OCS deployment engagement, but I'll try to give you some concise answers.

     

    1.  Microsoft's documentation generally assumes that you have a private WAN link beween sites. This is commonly acheived by using things like MPLS connections, but VPN is another option (though there are increased risks that need to be accounted for).

     

    2.  The Director looks up the SIP URI in Active Directory, finds the pool associated with that user, and directs the traffic accordingly.  This is identical to how users communicate between sites - remember that the Director is just a regular OCS server that doesn't contain users.

     

    3.  This is a common discussion item.  You can architect that very basically by pointing everyone to a Director in your headquarters or be as complex as using location-based DNS referrals through something like an F5 solution.  The drawback in the first example is a single point of failure while the latter example is expensive and more complex.

     

    4.  OCS is usually centralized except in very large delpoyments.  This is not a problem for IM, but the A/V and Web Conferencing experience can be negatively impacted by this (especially A/V) if there is a large amount of latency or low bandiwdth between sites.  It is not possible to place A/V and Web Conferencing servers by themselves in a particular location - you must deploy an entire pool.

    Thursday, March 6, 2008 1:54 PM
    Moderator
  •  

    THanks for clearring that up.
    Thursday, March 13, 2008 12:08 AM