First in a series of thoughts this month on IT, Cloud, SOA, bicycling, well, just about whatever...
--------
The new year is underway, party hats are put away, the confetti as been swept up. Now is as good a time as any to tackle some big problems. And large data sets can be a challenge.
In the cloud, a plethora of new data stores are coming on line, but challenges remain on how to scale (to accommodate upwards of gigs of bits), and how to distribute -- to the right user. Some areas I recommend to my clients for consideration include:
How often is the data accessed? Is caching the right way to go for performance reasons, or is this transactional data, requiring frequent Create, Update or Delete actions? RESTful services that map the traditional WS* approach to GET, POST, and other HTTP actions might help with performance (and portability of code).
Perhaps there is one (or more) enterprise application currently utilizing that big data set. Operating across platforms, perhaps even networks, raising different challenges. Are your enterprise apps shunting computational processing off to the database? Is that hurting others' use? One hint: when you audit your business processes, is business logic captured as stored procedures?
Shorthand solutions often present the quickest wins — keep data near the end user, for example. With mobile platforms (iPhone or iPad clients, for example), perhaps storing a subset of data on the device, and synching, makes the most sense.
Whatever big data set challenges you face, the approach to crafting solutions may seem templates. but crucial architectural strategies are pretty much unique to the specific situation. A major change to tables, or the question of meta data, can involve architectural changes on a large scale. And the new year is a good time to plan major renovations…
No comments:
Post a Comment