Customer 1

London based design office with outlying offices, now approximately 100 users.

 

Background

We originally migrated this customer out of a large organisation onto its own systems in 2005 with 25 users.  We helped maintain and update the servers with just disk upgrades until 2010 when they had grown to 80 people.  The servers and systems were at capacity and 5 years old in need of a refresh.  In addition the customer planned an office change with a new server location.

Virtualisation Customer Example

Partners

Design and Installation

After some research and consultation we designed a new infrastructure with the objective being able to cope with the demands of the next 5 years and beyond.  We designed a new infrastructure comprising two Citrix XEN nodes and a 3TB Scale Computing storage cluster.  The capacity of the 2 Xen nodes was deliberately specified so that all of the virtual servers (currently 5) can run on one node.  For normal operation the load is shared across the 2 nodes.  This provides the level of resilience required against hardware failure.

 

XEN Node Hardware

We specified and purchased two identical servers from RackServers.com as we were able to configure the servers to suit the customer requirement.  Additional XEN nodes can be added at any time in the future if the computing requirement increased.

 

Scale Computing Cluster

Given the remit to provide an infrastructure for 5 years we chose the Scale Computing storage cluster.  While the current storage need was 1 TB the rate of growth of data was unknown.  In this basis we settled on the 3TB cluster in the knowledge that expansion would be easy in the future by purchasing a new node and this provided the lowest start cost.

 

The scale cluster offered the simplest solution to install and manage and provide the right level of resilience against hardware failure.  The scale cluster comprises 3 nodes, any one hard drive or node can fail without interrupting service to the user.

 

The business servers are stored inside the cluster and can be booted form any of the XEN nodes.  IN addition because the servers are identical the servers can be moved to the other node while running.

This is a relatively simple and stable computing demand so moving the servers is a rare requirement leaving the servers to be managed as standard Windows servers.

 

Installation and Migration

After a protracted decision making process the XEN nodes and storage cluster were installed in April 2011.  The Windows server migrations and upgrades were almost transparent to the users.  The only minor issues that arose were down to reconfiguring the data stored on the drives to make it easier for the users.

 

Operational Experience

Since the installation of the nodes and cluster the Windows servers have only been rebooted as part of the normal operation and maintenance regime.  The storage cluster has run continuously, however a failed hard disk was replace after 3 months.  Where support has been requested form Scale computing this has been given quickly and remote access covered all the requirements.

 

Backup and provision for DR

While the infrastructure provided a highly reliable and resilient computing platform there is still a need for backup with provision for disaster recovery (DR).  Onsite backup is required to cover user error and any possible corruption of software or data, with an off-site copy for DR.

 

Image backup is employed through StorageCraft ShadowProtect software. Each server has incremental images stored on a backup computer.  In this case the hourly incremental images are collapsed into a single daily incremental image which is exported off site to a data centre every night.  The images, both on site and at the data centre, are periodically checked and powered up through Virtual Boot to both ensure their integrity and practise the initial disaster recovery process.

 

Conclusion

Customer 1 has a simple and reliable infrastructure that has enabled everyone to continue normal business activities inside and outside of the main office reliably and without worry.  The data growth has exceeded the initial projection as another node is now required for additional data storage.

 

 

@ 2013 Proven Virtual Servers c/o Infosysco