THE SQL Server Blog Spot on the Web

Welcome to - The SQL Server blog spot on the web Sign in | |


You searched for the word(s):
Showing page 1 of 55 (544 total posts) < 1 second(s)
  • re: SQL Server on Xeon Phi?

    the Xeon Phi is not cheap. It may be cheap to build a chip with 4-8 Atom cores, building a giant die is not. Networking cheap computers only works for workloads that can tolerate high latency. As I said above, the nature of Moore's Law is that a complex core does not have linear gain with the number of transistors. Some workloads really need a ...
    Posted to Joe Chang (Weblog) by jchang on August 30, 2016
  • re: SQL Server on Xeon Phi?

    The objectives for the current Xeon Phi is proof-of-concept. That the smaller/simpler Atom core at 1.3GHz are not far off the powerful Broadwell cores at 2-3GHz+ in transactional queries, with the expectation that Broadwell at 3GHz+ are exception for code running largely inside L2 cache. Also, for transactions, SMT should be more than 2-threads ...
    Posted to Joe Chang (Weblog) by jchang on August 24, 2016
  • SQL Server on Xeon Phi?

    Can the Intel Xeon Phi x200, aka Knights Landing, run SQL Server? It does run Windows Server 2016, so is there anything in SQL Server 2016 that would stop it from installing? Xeon Phi is designed for HPC, so it would not have been tested with SQL Server, but that does not confirm whether it will or will not work. If so, then this could be used ...
    Posted to Joe Chang (Weblog) by jchang on August 23, 2016
  • re: Server Sizing Comments: the case for 1-socket in OLTP

    SQL Server 2005 forward may try to schedule all threads on the same node but does not guarantee it. I do remember situations where 7 threads were on one node and the 8th was on another. I am of the opinion that when MAXDOP is set to N, N being the number of physical cores on one node or less, that the DOP strategy should be to give me up N with ...
    Posted to Joe Chang (Weblog) by jchang on April 14, 2016
  • re: Server Sizing Comments: the case for 1-socket in OLTP

    no I have not read it until you pointed it out. I do know that older versions of SQL Server, 2008, not sure about R2, that a parallel execution plan might get some threads on one socket and some on another when MAXDOP is less than the total number. But back then, perhaps MS was not fully up to speed on such strategies. It does seem that 2008R2, ...
    Posted to Joe Chang (Weblog) by jchang on April 13, 2016
  • Server Sizing Comments: the case for 1-socket in OLTP

    Long ago, almost 20 years, there was legitimate concern on whether servers had sufficient capability to drive transaction processing requirements for large businesses. Then for a long period, perhaps from the late 1990s to 2012, the better choice for most mid to large businesses was to simply buy a 4-way (socket) system. Larger systems were ...
    Posted to Joe Chang (Weblog) by jchang on April 12, 2016
  • re: PDW Concept on NUMA

    it has been a long time since I have looked at PDW, is it APS now? I lost interest when they went 7.2K HDD only. I think MS has become dominated by marketing types who become enamored with paper specifications, in this case, the size of the database, rather than how it is used. I am of the opinion that there are situations in which we want massive ...
    Posted to Joe Chang (Weblog) by jchang on January 7, 2016
  • re: SSD Form Factor and interface

    My opinion is that the cloud is a response to IT infrastructure practices. Too many buy gold bricked systems, mandate standard configurations that do not suit databases etc, implement procedures that obstruct the ability to do information technology. There have times I have spent several weeks trying to negotiate with the infrastructure team on ...
    Posted to Joe Chang (Weblog) by jchang on December 29, 2015
  • re: Data Tales #5: The Case of the Rogue Index

    hello Greg Not that I think it is a smart idea to disable row locks willy nilly, but if you want to speculate why, in my slide from CMG2004, slides 50-53 in query performance of the key lookup operation versus rows, there is a major collapse in the neighbor of 3000-5000 ...
    Posted to Greg Low (The Bit Bucket: IDisposable) (Weblog) by jchang on December 17, 2015
  • re: Computers without ECC memory are crap - no exceptions

    thanks donaldvc, I will read this when I get a chance. That ECC for single/double bit errors is necessary has be known for a long time. From a quick skim, I think this article is stressing the importance of correlated multi-bit errors, given the organization and high density of DRAM. My comments will be a the bottom of the main post.
    Posted to Joe Chang (Weblog) by jchang on December 10, 2015
1 2 3 4 5 Next > ... Last »
Privacy Statement