= Сообщение: 7471 из 12496 ====================================== ENET.SYSOP = От : Gerrit Kuehn 2:240/12 25 Sep 18 21:40:26 Кому : Nick Andre 25 Sep 18 21:40:26 Тема : Re: RBB FGHI : area://ENET.SYSOP?msgid=2:240/12+5b620c70 На : area://ENET.SYSOP?msgid=1:229/426+C10644F1 = Кодировка сообщения определена как: LATIN1 ================================= ============================================================================== Hello Nick!
25 Sep 18 14:48, Nick Andre wrote to Gerrit Kuehn:
GK>> However, in non-virtual setups every server can only fail the GK>> services it i running. vmware server will fail everything they run.
NA> If your work cannot afford redundant servers or infrastructure, then NA> that is NA> not a Vmware problem or a server problem.
No, this is not a vmware problem, it is a general virtualisation problem: To achieve the same stability, security and deliver the same service as with traditional servers, you'll need much more than the virtual server initially advertised. A few weeks ago, colleagues had their virtualised single sign-on failing on them, taking down *all* services needing it. The virtualiser was redundant, storage was redundant, but still didn't help. It took them 5 people and 36h to get back online.
Lesson: making systems bigger makes the issues you'll have when they break even greater. Putting in more redundancy just makes you wait longer for a desaster bigger than anticipated. However, it will always happen, you'll just have to wait long enough. When it happens, your big system will fail epically and produce more chaos than you've ever seen before. OTOH, a larger number of smaller systems with dedicated services (preferrably deployed in a decentralised way) will fail more often, but these failures will be smaller, easier to handle and have less impact on the rest of your systems. You choose.
