Virtual Infrastructure
This month in the virtual infrastructure, we enabled HA and DRS across the entire virtual environment. HA (High Availability) is a VMware technology that automatically powers on VM on available resources if the host they were running on faults. DRS (Dynamic Resource Scheduling) is a VMware technology that constantly evaluates a ESX Cluster. If monitors the hosts and VM’s in the cluster, and runs calculations to ensure that all VM’s are getting as many resources that the cluster can make available.
We also identified a need to manage the capacity planning of the virtual infrastructure. We put together an excel spreadsheet that with minimal manual input trends the capacity of the virtual infrastructure. Since we just started gathering historical data last month, it will take a few months to see the trend. However, we were able to immediately identify where we needed more computing resources to provide an efficient, stable, and highly available virtual environment.
We also made the virtual hosting service a campus available service, that departments can pay a small charge for virtual resources. We developed a feasible SLA (Service Level Agreement) around this service to set the expectations of customers.
List Serve Evaluation
We are looking at life cycle replacing the list server, and we are looking into different software options. We have identified a product that may server our current functionality as well as provide some additional features. We want to implement these servers on virtual servers. We have received specs for the hardware and created a VM to evaluate this product on.
Navigator
We finally have two navigator servers. We did many test of failover since NEC has no option for redundancy with this application. So we wrote a sync script that will synchronize the configuration directory of the NEC application on both servers. With this enabled, all we have to do is move the licensing dongle from one server to the next, make a MYSQL update, and start the navigator application. This is the best we can do at redundancy for this service.
Backup Server
We replaced the SUN Solaris server we were using as our backup server with a Linux server. There were special storage, and networking requirements that we had not implemented before within ITS applications hosting. These include: OS layer multi-pathing on HBA’s which allows redundant connections to SAN storage, IEEE 802.3ad with VLAN tagging which allows multiple VLAN (networks) and coupling of NICs together (we couple three NICs in one 3GB bond). There have been issues since this migration however, the Linux and the storage team has been working diligently to mitigate these problems.
IDM
We are gathering information from future customers of this service in order to ensure we are going to meet their needs. We have made many decisions around this space, yet it is still in preliminary stages for applications hosting involvement. We have been making recommendations to the implementation team around architecture and strategy, which will lead to the design document. This is very important because the implementation will be based upon the design document.
CSM – Load Balancing services
We are looking into making the proofpoint appliance mail firewall a load balanced service as it will simplify scalability and also increase reliability. We put three proofpoint test appliances behind the CSM load balancer module. We then scheduled a test of 30,000 message / hr (very similar to production load) to verify the functionality of the CSM in front of the proofpoint mail appliances. The CSM performed extremely well, as it was removing nodes that had stopped responded to SMTP request because of load, and putting them back once the SMTP service was available again.
We are implementing a sharepoint environment, and stretgically this will need to be a highly available service, so we putting these servers behind the CSM. We have been working closely with the Windows administration team to provide them with CSM configurations that will assist in scalability and availability of this service.