I work in a company, where I was lately having discussion with folks from system engineering and the security team on topic of fast data access in terms of (IOs) performed over the network (SAN) versus local disk storage as to who is faster then the other and for what reasons.
The key requirement in our case, was not actual storage but the first recovery of data for analysis like in case of SIEM technology.
The guy from system lead was insisting , that he runs all of his BLADE infrastructure by booting it from SAN instead of local storage and says he has no problem in performance of those servers.
I have studied computer engineering back in college, and what little I remember or recall is that bus-speed calculation is usually fast enough, and slower only to memory in access time.
I don't know about the finer details of disk manufacturing but is this issue of disk IOS subjective in nature and the disk seeks times varies from vendor to vendor or is there any scientific model or concepts that goes deep down to the design of electric circuits that prohibits a SAN storage to be faster or slower to disk storage and retrieval.Also on what levels can should we compare the two together e.g RAID configuration to be equally set.
Also, other interesting thing that I had in mind was how the big companies like google or youtube performs data analytic and data mining for business intelligence do they not query SAN. Or they can very well be running huge clusters with internal storage / disk.
I would appreciate if anyone can help me guide on this subject that bring some scientific data in the picture for this discussion.
In case you need a background on technology I'm talking about. I'm talking about ibm storwize 3700 storage and local disk of specs / configuration of RAID 10 rpm of 10K. In terms of data I'm talking about 60 GB of daily growth.