Yet another sign that the Register jumped the shark
In ancient times (in IT world: 2005 or so) the Register was a good source of informations. You´ve got interesting hints to technology and a realistic view to IT, the major players and the newest hypes. Several years passed and now the Register is totally unreadable. I just keeping it in my feedreader as it´s important to me to be aware of Register articles before customers confront me with the nonsense written at the Register.
One of the names connected to the downfall of the Register (at least in my opinion) is Chris Mellor. Most of the time i ask myself about his connections to NetApp, but one of the best examples to give you an impression about the questionable quality of his articles is his recent output regarding about the Oracle ‘faster, cheaper’ with VMware. This article is so bug-infested, that i doubt, that Mr. Mellor did even basic research on this article.
To state it at the beginning: I´m still strongly opinionated that Oracle RAC isn´t a good idea to scale workloads, but it´s a great way to shorten your Oracle installations failover time. This is my opinion and this opinion won´t change … even in the light that the logo on my business card may be red in the future ;) Due to the nature of RAC it solves the problem of the need to replay the logs of your database after failover. This is a problem that strikes you with normal HA clusters. Dependent on the load on your database there may be enough “to-be-replayed” transactions in the database to hold up the startup of your database for several minutes.
What´s the point of Chris Mellor in his article: Well, he tries to pitch the use of VMware as an failover technology. He does it in an intelligent way. Many sentences begin with “Is said to”. Journalism is more than paraphrasing articles and prepend the sentences with “Is said to”.
The problem is: The foundation of article is based on a blog article of an EMC guy. You know: EMC is the company that purchased VMware a while ago.In order to pitch VMware HA the referenced article omits many facts or represents them in a wrong way.
For example: He is stating in
both Oracle Cluster Ready Services (the underlying technology behind Oracle RAC) and VMware HA cluster are cluster software products.
This is highly misleading because the core of RAC is the Cache Fusion, the mechanism to provide a single SGA in a cluster. Cluster Ready Services or Sun Cluster (on Solaris you use Solaris Cluster for clustering for e.g.) is meant to make the infrastructure beneath RAC high available, as Oracle RAC assumes, that the system is made highly available. But the product isn´t bound to Oracle RAC as you you could see in the respective licensing rules of the Oracle software. By the way: I find this a little bit strange. You could do such a failover with a decent cluster framework on the native operating system without paying the VMware tax. I don´t talk about the license price of VMware, as you have to pay for a cluster framework as well (albeit not as much as for a VMware infrastructure). I talk about the loss in performance and the increase in complexity by moving a virtualisation layer between the native hardware and the database. When you wan´t clustering without RAC just purchase a Sun Cluster License with an Agent for Oracle or the respective Solution of Veritas. Or when you want to use an application specific cluster, you could use the Clusterware cluster, as this product isn´t bound to RAC. In the licensing information of Oracle 10g R2 you find the following rule:
Oracle Clusterware can be installed and used to protect any Oracle or third-party software provided one of the following conditions are met:
1.The software being protected is from Oracle.
2. The software being protected uses an Oracle Database.
3. The software being protected is running on Oracle Unbreakable Linux.
4. The software being protected is running in a cluster where at least one machine involved in the cluster is licensed using the appropriate metric for either Oracle Database Enterprise Edition or Oracle Database Standard Edition.
This is the point where the arguments of Mr. Mellor and the blogs he has linked to in his article breaks apart. In short you don´t need RAC for clustering, you don´t need the Enterprise Edition for clustering. Even when you want Oracle Clusterware. This information was just a single Google-search away. Furthermore: As far as i understand the HA cluster mechanism in VMware it´s … well … it´s just really basic. For example clustering the virtual machine means it´s totally unaware about the application running in the VM. It just checks, if the hardware running the VMware is still available. And that´s just a really basic functionally. It can´t restart your oracle in the case it had a hickup blocking the listeners for example. It can´t even detect this situation because it doesn´t check this situation, it doesn´t have the mechanisms to check this situation. The solution pitched by Registers Chris Mellor and the EMC bloggers is not unsimilar to an containerized Oracle in conjunction with the Zone HA agent in Sun Cluster. But this even this really simple solution would be superior to a VMware HA solution as the container HA agent can look into the zone and monitor the components in the zone, thus monitoring the status of the application and not just the status of the hardware. Of course you need some know how about enterprise computing and clustering to notice this subtle but very important differences. I would like to read articles at the big news outlets showing they were well researched. The article in the Register leaves this wish unanswered, as it doesn´t show a single sign of real know how, a single sign of journalistic research. To be honest: After all the article i´ve read from him, i don´t have any hope to read an decent article written by Chris Mellor. To end this article with a small wish: Mr. Mellor, it would be nice, when you bite the hand of EMC instead of simply copying an EMC blog article.