“Something old or new, borrowed or blue” - forgive the paraphrase of the items chosen to bring good luck to brides! Recently I’ve been involved in several situations where SAS customer IT organizations have proposed architectures for new SAS software that are not up to snuff. It occurred to me that this familiar wedding rhyme is also an apt description for software environments that are cobbled together without proper planning.
One customer is building a sandbox to try out a new implementation of SAS® Visual Analytics. This exciting visualization software is intended to fly – to bring graphs, tables and forecasts, based on vast stores of data, to your laptop, tablet or smart phone screen fast, fast, fast! And when the infrastructure is sized and set up appropriately, it will. But if it isn’t, it won’t.
Another customer has licensed SAS® Enterprise Miner, SAS® Text Miner, and SAS® Forecast Server – high-impact, advanced analytic software that uses a lot of processing power, a lot of data and extensive I/O capacity. The SAS users here are very excited about getting started with these solutions, but, again, the infrastructure slated for the software is insufficient to handle the number of users, the quantity of data and the type of jobs the users will be running.
Do the Planning
In both of these examples, the IT organizations assembled hardware solutions that ensured an acceptable bottom line: some of it old, some of it new, some of it borrowed, some of it blue, with the good intention of repurposing hardware for the new software. But will it ensure a good customer experience? Not necessarily. And certainly not in the long run!
Up front due diligence helps ensure that all the pieces are in place: the SAS application servers have sufficient fire power, the metadata and web tiers can support the load they’ll be handling, data storage is sufficient to enable a lot of incoming and outgoing data movement, and the piping (networks and IO channels) is big enough to allow for smooth transfers from here to there.
Use the Resources Available to You
How can you ensure that there will be a happy marriage between the hardware and the software, supporting contented SAS users? There are several ways:
- Make sure that the business users and IT are both involved in discussions with the SAS sales team and technical architects as early in the process as possible. This way, everyone has a chance to contribute to the conversation, ask questions and have a role in properly setting expectations around usage and support.
- Ask your SAS sales team and technical resources to have a sizing estimate done for you. SAS’ Enterprise Excellence Center (EEC) has a sizing questionnaire that is tailored to your specific set of SAS software. The questionnaire addresses such issues as the overall number of users who will be working with the SAS products, the number of SAS sessions executing concurrently at any given time, the type of processes performed (data manipulation? data querying? light reporting? advanced analytics?), the typical amount of data used in these different sorts of processes, and relative usage at peak and non-peak work hours.
- Use the ECC results to guide your purchase of new hardware or to gauge the feasibility of using hardware you already have (The EEC will provide you with a hardware estimate based on the questionnaire responses). If you were hoping to utilize something smaller than what is recommended, that’s fine – you can scale down the number of users and processes for now to maintain optimum performance, and can scale up later as the word spreads about how awesome SAS software is!
- Make use of a wealth of documentation and papers to guide you toward a successful implementation, including the following:
- SAS System Requirements
- How to Maintain Happy SAS®9 Users
- Frequently Asked Questions Regarding Storage Configurations
- Best Practices for Configuring your IO Subsystem for SAS®9 Applications
- SAS Deployment: do you need an architectural plan?
- Ask your SAS sales team or your Customer Account Executive for additional architecture and performance papers.
Ensure a Happy Ending
Everyone wins when the SAS environment is properly sized and tuned. The SAS users can get their work done in a reasonable and predictable amount of time. IT deals with fewer trouble tickets related to backlogs and performance issues and avoids trying to making an ill-fitting environment fit better.
We all know that everything won’t always work smoothly and that tuning, adjusting, fixing, and growing the environment are part of the game. But with appropriate planning up front, a lot of the worst problems might be avoided.
I’d love to hear your examples of how planning and preparation led to a happily humming SAS environment!