These days, with IT budgets stretched to the breaking point, you probably find your company routinely keeping your servers for 4-, 5- and even 6-years. Pretty much gone are the days of replacing a server that’s still performing relatively well after three years. Whether to maintain a current business application for another 9- to 18-months, or to repurpose server hardware for a new role, you may be considering upgrading your existing servers. Sometimes this can be a good choice, but in other cases this could be a bad business decision. This blog article is to help you evaluate your different options, allowing you to make a good business decision.
Applications today demand more processing performance and memory capacity. Intel’s tock/tick microprocessor architecture and socket compatibility has allowed the last five server generations to support processor upgrades that wouldn’t have been possible before. Memory, which has recently seen an unprecedented price increase, is still relatively inexpensive, and allows servers that might have started with only 4GB or 16GB to grow to larger memory sizes. Larger capacity drives available today allow for storage growth. So, what are the pros and cons of upgrading older servers rather than buying new ones?
Related: Ditching the Hardware: An Introduction to Serverless Computing
Upgrade—Pros
- Some increased performance and/or capacities, for a reduced additional investment
- Possibly a quicker implementation, with less data center disruption than a server refresh
- Avoids purchasing new server licenses (iLO Advanced Pack, OneView, operating system)
Upgrade—Cons
- The application is running on hardware with support that is about to, or has already, expired, and the cost to extend support another year is not insignificant
- Less system performance than a current server generation would provide
- Less memory capacity than current server generations, plus more expensive memory
- Other system components are previous generations/versions (array controller, network adapter, SAS drives, PCIe slots), widening the performance difference
- The server is reaching the point where you might start to see some hardware issues pop up due to the server’s age
So when does upgrading make the most sense?
A conservative memory upgrade is something that’s easy to do and relatively easy on the budget. I say “conservative memory upgrade,” as I had a customer that needed multiple servers (with 128GB to 256GB) upgraded to 1TB, and they were shocked by the cost. In most cases, all the existing memory had to be pulled (servers can’t mix RDIMM and LR-DIMM), and the memory for their older v3 processors was significantly more expensive—so much so that they could upgrade the processors to v4 (which used less expensive memory), add 1TB memory, and still save money! This is a point where a processor upgrade actually helped; it was still a very expensive upgrade, and in the end, they upgraded some servers, but purchased a few new ones, as well.
Information that your systems engineer will need to recommend the correct new memory modules are: server make, model and generation; specific processor model and generation; details on the current memory installed (quantity, size, and type). [Ex., HPE DL380p Gen8, E5-2620v2, 8x 8GB RDIMMs]
Often this information can be found using the server management processor (e.g., iLO, DRAC). From that, your systems engineer can plan to re-use as much existing memory as possible, minimizing the upgrade cost.
Adding additional drives or replacing existing drives with larger capacity drives or solid state drives is another common upgrade request. You need to evaluate if the older server can be upgraded to the new storage capacity and if the SAS speed (3Gbps, 6Gbps) is acceptable. Later server generations often can take greater drive quantities, so an upgrade on a server that can hold fewer drives would mean purchasing larger capacity drives, which are often more expensive and will incur a greater RAID loss. These cons can sometimes be negated if additional drive cages and controllers are still available for the server you’re considering upgrading, so this type of upgrade can work out well. Your systems engineer will need similar information on the current internal storage configuration to properly advise you on the available options.
In most cases, upgrading the processors can be a fairly expensive upgrade, and if the memory is not upgraded at the same time, the system won’t see the full benefit of the new processors. I suggest asking a Connection Account Manager to price out both the upgrades and a new replacement server, which might provide enough additional performance to allow a 2:1 or 3:1 consolidation. When the numbers work out in favor of a new system, it’s easy to justify replacing an older server (or servers) with a new one that provides faster performance and full warranty/support.
Upgrading can be the more expensive choice
When these are business-critical applications, downtime can be very expensive. Adding an additional year of support to an older server to protect the business could be the cost that pushes the decision toward buying new servers.
The pros and cons for buying new servers rather than upgrading are mostly a reverse of the pros and cons for upgrading shown above:
Buy New—Pros
- New server hardware can come with three, four, or five years of warranty or support coverage
- Higher system performance than the older server would provide, which might allow 2:1 or 3:1 server consolidation
- Higher memory capacity than the older server would support, and likely less expensive memory
- Other system components are latest technology, providing higher performance
- Other than unlikely manufacturing flaws, a new server should not see any hardware issues
- Longer expected life of a new server
Buy New—Cons
- Likely a higher new investment, but with the expectation of using that server for a longer time
- Typical data center disruption for a server refresh
- Server licenses (iLO Advanced Pack, OneView, operating system) often need to be re-purchased
As mentioned earlier, pricing out both upgrade and new options can help you make a good business decision. Connection can work with you to look at all your options for upgrading or replacing your server infrastructure, ensuring that you’re making the best value decisions possible.
Jim Fields
Jim Fields is a Senior Systems Engineer at Connection with more than 38 years of experience in servers and server management. He holds several technical certifications, including HPE Master Accredited Solutions Expert (MASE)—Advanced Server Solutions Architect V3, HPE Accredited Solutions Expert (ASE)—Composable Infrastructure Integrator V1, and HPE ASE—Server Solutions Architect V4. In his free time, Jim enjoys playing the trombone and singing at his church, as well as computer programming and volunteering with the Boy Scouts.
