The solution you get is only as good as the problem you posed!!

The point is, I believe the only thing that we should help our clients do, is to help them frame the correct problem statement.? If this is done, then we all eventually reach the correct solution and hence correct outcomes, which lead to successful engagements.

Right now, there are plethora of products in the market place. for example, consider All Flash Array. There are about nine offerings from IBM alone, and multiple more in overall storage market with all put together. ?Prima facie all of them deliver high performance, low latency using SSD/Flash cards with some form of data reduction. Then how do we choose? If all are same and business needs are same then why are some clients successful with one product and others are not??

More often than not the over simplification of problem statement leads to sub-optimal solution choices. We need to know exact problem and expected outcomes, most importantly in order of priorities, so that we can optimize without losing the end objective.? Some key questions can be

?1. What is that we gain if a query responds few microseconds faster??

2. What is the cost of delay of say 500 microseconds for IO response?

?3. Is the performance driven at compute OR at IO? What is the next bottleneck part from IO? If IO problem is resolved, will the bottleneck shift and limit the gain from faster IO??

4. What kind of data is it??

5. Will data reduction result in significant savings??

6. Does the saving with data reduction outweighs the cost of drop in performance?

?7. Does the data need to be encrypted? Does the penalty of encryption on performance a reason for not implementing encryption??

8. Does the data need to be replicated? Does the latency of replication impact production performance? Is there a way to isolate the same??

9. Does failover and fallback from remote site need to be automated?

These are really few of indicative questions, I am sure technical folks reading this post can think of many more. However, the idea is to form a consensus, evaluate all dimensions and formulate the problem statement that captures stated and more importantly unstated but assumed needs and expectations.? Clarity here can make many decisions look straightforward and help us arrive at a naturally fit solution. ?For example, following is a crude chart on decision making tree with IBM flash systems which can lead us to a single product based on decision criteria and reveal glaring gaps in problem statement and understanding.

Good solution and Good solution delivery are only result of precise problem statement. I believe that’s the key in all technical resolutions and ultimately client success and satisfaction, pre or post sales!!

This blog was first published here.