Close this search box.

Stop software upgrades killing the network


The jury is still out over whether Windows 8 will be a hit for enterprises due to the radically new user interface and the perceived lack of new features from previous versions, but it is likely that a fair few network administrators are dreading the day when the upgrade is rolled out. Technical Product Manager for enterprise systems management at 1E, Richard Threlkeld, explains why software upgrades impact network performance and discusses how to minimise disruption.
The need to support an ever increasing number of enterprise applications has always presented a challenge to network administrators. At no time is this more acute than when Operating Systems (OS) upgrades are rolled out. With the recent release of Windows 8, networking professionals will be once again wondering how to provision these important software upgrades without choking the network and, even more crucially, without negatively impacting business critical applications.
The root of the problem lies with the fact that IT software content is continually growing larger, whether this relates to user applications, server applications, system and software updates or OS content. Indeed, typical OS content, which includes core images, boot images, driver packages and software packages, has multiplied in size in recent years, so while a base Windows XP image (after performing a system capture and including only the service pack) is around 800 MB in size, the equivalent Windows 7 Enterprise x64 image is approximately 26 GB.
From a networking perspective the challenge is not just about the growing amount of data that needs to move from one location to another. It is complicated by the fact that both WAN and LAN speeds growing at a much slower rate than content is increasing, making it a mistake to try to push out all data as fast as possible across the network. This approach will only result in a situation where it takes a long time to deploy content and – if not managed correctly – congestion.
Even if the event the network does have spare capacity, applications will still compete for any available bandwidth, so it is vital that business critical applications are given priority over IT usage. However, there are also many critical projects that IT needs to complete in order to enable the business not only to function but also to advance from an operational standpoint. Some of the highest priority IT projects are OS deployments, but all too often these are very costly and time consuming to perform, not only from an application compatibility perspective but also when trying to transfer content to remote locations or when automating the migration process. To avoid clogging up the network, it has become important not to shuffle user data across the network during the migration process, while it is also preferable to avoid having to deal with network configuration or traffic when doing Preboot eXecution Environment (PXE) requests for bare metal system builds.
As it is inefficient to maintain a network with excess capacity, and impractical to add extra bandwidth to support ad hoc IT projects, many enterprises have opted to deploy application management solutions or bandwidth compression tools. While these can improve network efficiency, they are by no means perfect solutions, adding management complexity and merely addressing the symptoms – rather than the cause – of the problem. To truly alleviate the congestion caused by software upgrades, the roll out process itself must become more network-aware, automatically knowing when to deploy and, just as importantly, when to back off.
Software configuration platforms that look at how data packets traverse the WAN and regulate bandwidth using techniques such as autonomic throttling solve the network congestion problem at source. In addition, by automatically holding local or branch office elections to determine a single desktop which will pull content across the WAN, they significantly reduce wide-area traffic, thus speeding up deployments and reducing the risk of network outages. Similar efficiencies can also be achieved in the LAN, with each branch’s representative desktop only pushing out upgrades to other PCs when it knows no business critical applications can be impacted.
If these are the kind of issues that you have concerns about or are experiencing, Nomad 2012 is one of these types of intelligent solutions that can help alleviate these problems.


The FORRESTER WAVE™: End-User Experience Management, Q3 2022

The FORRESTER WAVE™: End-User Experience Management, Q3 2022