From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-1.9 required=5.0 tests=BAYES_00,FREEMAIL_FROM autolearn=unavailable autolearn_force=no version=3.4.4 X-Received: by 10.68.92.4 with SMTP id ci4mr2325842pbb.7.1378375921973; Thu, 05 Sep 2013 03:12:01 -0700 (PDT) X-Received: by 10.49.104.83 with SMTP id gc19mr36205qeb.12.1378375921761; Thu, 05 Sep 2013 03:12:01 -0700 (PDT) Path: border1.nntp.dca3.giganews.com!border3.nntp.dca.giganews.com!border1.nntp.dca.giganews.com!nntp.giganews.com!n2no22306955pbg.1!news-out.google.com!z6ni33820pbu.0!nntp.google.com!j7no294380qai.0!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail Newsgroups: comp.lang.ada Date: Thu, 5 Sep 2013 03:12:01 -0700 (PDT) In-Reply-To: <87txhzu13v.fsf_-_@adaheads.sparre-andersen.dk> Complaints-To: groups-abuse@google.com Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=105.236.58.84; posting-account=p-xPhAkAAADjHQWEO7sFME2XBdF1P_2H NNTP-Posting-Host: 105.236.58.84 References: <9fc26b8a-7a5c-4941-81cb-8f0e9c17a660@googlegroups.com> <2383e534-646c-4500-ad05-40caf9460c35@googlegroups.com> <87txhzu13v.fsf_-_@adaheads.sparre-andersen.dk> User-Agent: G2/1.0 MIME-Version: 1.0 Message-ID: <4362f3e7-ea19-4054-84e2-33e87a1bb572@googlegroups.com> Subject: Re: Memory mapping and network file systems (Was: B-tree performance) From: Peter Brooks Injection-Date: Thu, 05 Sep 2013 10:12:01 +0000 Content-Type: text/plain; charset=ISO-8859-1 X-Original-Bytes: 2071 Xref: number.nntp.dca.giganews.com comp.lang.ada:183307 Date: 2013-09-05T03:12:01-07:00 List-Id: On Thursday, 5 September 2013 10:25:08 UTC+2, Jacob Sparre Andersen wrote: >> > If the persistent objects are shadowed in RAM disc and that's shared > > over a network, it can make the solution nicely scalable across > > multiple networked machines. > > That sounds rather dangerous to me. I would expect either very low > performance or lots of data corruption. > -- Robert I. Eachus > So would I. Unless, as is the case, triplestores are more read than written, so only one process needs to write, removing the main source of corruption and removing the need to do lots of locking.