Site Survivability
,For Lync Server to better compete with other enterprise-grade telephony platforms, resiliency and survivability needed to be addressed. New, additional options for Lync Server topology, along with the introduction of new hardware and server roles, coupled with the support of new DNS options, provide Lync Server architects the ability to craft deployments that are highly available and able to survive failures at various points in the enterprise. This enables Lync Server to continue to provide vital telephony services to Lync Server users.
Survivable Branch Appliance
With the registrar role moved to the Front End Servers and possessing its own SQL express database, pools now have reduced requirements on the back-end SQL database. A survivable branch appliance (SBA) can be set up for branch users as their primary registrar with the pool as their backup registrar.
Lync Server branch users still get their user services from the Front End pool, usually located in a central datacenter. However, in the event of a pool failure, because the branch appliance is aware of the branch user registrations, users at the branch will experience only a loss of user services and still be able to access the PSTN because routing is running on the SBA. Unlike some traditional PBX branch scenarios, Lync Server users benefit from this topology change by not having to re-register to the SBA during a failure.
Supporting DNS Load Balancing
By supporting DNS load balancing (DNS-LB), enterprises that deploy Lync Server can benefit greatly thanks to simplified hardware load-balancing (HLB) configurations. In enterprise HA deployments, HLB are still required for certain traffic, notably HTTP and HTTPS. However, because these are the protocols that are commonly run through load-balanced configurations, their deployment is simpler than in previous versions when SIP traffic also passed through HLBs.
Note
The use of DNS-LB allows for simpler server shutdown through draining. This is a benefit that any support engineer who has ever had to take a server out of service can greatly appreciate! In an N+1 scenario, where a subset of the servers in a pool can support the entire enterprise, it is possible to remove a server during normal business hours.
With Lync Server, SBAs are now managed from the CMS database, which provides tremendous savings in the deployment and management of remote locations. Help desks and ISVs can prestage a branch appliance prior to shipping to a remote site. Once onsite, a technician can complete installation. This ease of deployment can be repeated for an unlimited number of sites, greatly reducing the workload of domain and enterprise administrators.
Another topology benefit is a new role, known as a branch office server that can support approximately 1000 users. The new role enables enterprises the flexibility to standardize their deployments across many branches of varying sizes, without sacrificing reliability, providing highly available PSTN connectivity.
In OCS 2007 R2, a common topology was to have dedicated pools at regional datacenters. With this new backup registrar capability, these same deployments can provide an available telephony solution by simply designating an alternative pool from another datacenter as the backup registrar. This feature is known as data center resiliency and provides a limited set of features, including PSTN access, to users whose primary datacenter is unavailable.
When datacenters are connected through low latency (<15ms rtd) WAN links, a single pool can be spread across multiple datacenters. In this configuration, an enterprise provides the entire robust set of Lync Server features out of either datacenter. This configuration is known as metropolitan data center resiliency.