For scaling CMS sites, it's often a case of compatibility with 3rd party tools.
My recent experience has been with Drupal, but the tools are the same for other *AMP based tools (ie. PHP/Apache/MySQL based), like most open source CMS's are.
The real key thing to determine for you will be the compatibility of each CMS with these standard tools for getting the scalability happening. That is almost more important than the specific scalability features of the CMS itself, which are pretty much the same in most modern CMS's (aggressive caching and so on).
The typical pieces of the puzzle for Linux are the following. You should evaluate which of these pieces matter to you and which will work with the CMS you choose.
Varnish: This HTTP reverse proxy will sit in front of your main webserver and efficiently serve up anonymous page requests where it can. If you have a site with lots of anonymous (not logged in) traffic, then Varnish will pretty much give you the same performance you'd get if your site was static HTML pages. The catch: Your CMS must cooperate with Varnish, either through an add-on module or direct knowledge of how to behave behind one of these proxies. Anonymous pages must be served up without any unique cookies on them, or Varnish cannot do its thing.
Memcached: This in-memory object cache is used as an alternative to cache tables in your SQL database. As we all know, memory access is 1000 times faster than disk access, so it stands to reason that if you keep these cached bits in memory, instead of having to use the database and spin the disks to run queries.. that your site will be very fast. It also stands to reason you need enough RAM to hold extra copies of stuff in memory.
Caching in the CMS: All modern CMS's try do to this as well as they can, it's the key to performance. The achilles heel of all modern CMS's is that you can't really cache logged in data very well (ie. a dashboard that has different stuff to say for each user). So if your users are all logged in, say for a community site or something, then neither this nor the two items above is going to help all that much.
CDN: When CMS people throw around "use a CDN" they mean a couple things. One is what it sounds like, host stuff like static files on a content delivery network provided by Google or Amazon or someone, and things will load faster. But the other thing it means is to set up multiple parallel DNS names to load content off. Your browser will load 6 things at once max from a single domain. If you make: a.example.com, b.example.com, c.example.com, d.example.com, e.example.com, and then if your CMS supports this CDN technique, then a browser could download 30 things at once instead of 6 - it could grab 6 items (.css files, .js files, images, whatever) off each of a., b., c., d., and e.example.com. This speeds sites up, but again your CMS has to support it (built-in or through some kind of add on module)
Database replication? This is always prickly if you are talking about doing it in MySQL. Regardless of database, replication is a tough problem to solve, especially if it has to be 2-way replication (each side has to read and write). If you do everything above perfectly, and buy the best hardware you can, and stick the database on its own high performance server with no other services on it, and a nice fast raid-10 array say with 6 disks to get yourself really awesome disk performance........ then you could probably go a long way with a single master database. (it would definitely be a wise idea to replicate to a backup machine if you had all your eggs in one basket like that). Anyway, by this point you are in standard database replication land... Your CMS might support having multiple database entries, but it's largely more a question of "how do you cluster" rather than "what CMS are you clustering". I'd be surprised if any of the modern alternatives wouldn't allow you to set more than one database hostname for clustering, and you could always do it in DNS by round-robin'ing the hostname of your DNS server. Again though, not related to your choice of CMS really.
And to end this all of, and to answer your question, I of course recommend Drupal as a rock solid scalable CMS. Is it easy to scale it so it runs a site like examiner.com or whitehouse.gov or a university site with hundreds if not thousands of sub sites and millions of nodes? Heck no! Nor will that challenge be easy in any CMS you choose.. but you should also be able to choose any modern popular CMS and make it work fast, if you address all of the above.
We are currently using Alfresco in a clustered setup with rather high loads. Works fine. We have noticed minor issues with lucene index synchonization though- it may take a minute or two for it to replicate through all nodes.
The GA version is open source. They also provide extensive documentation and training.
This is my opinion and some might not agree with it, but the majority of medium to enterprise content management systems (both open source and commercial) offer load balancing functionality. The question is how efficient and scalable is their solution. For example, Sitefinity offers load balancing where each server in a web farm must be aware of other servers. For example, if I have 20 servers in a web farm, then I have to configure each server and make sure it's aware of other servers. This is a real pain, but it's still a "load balancing" feature. What I'm trying to say is make sure that you investigate on how the load balancing works in a certain CMS before you decide to use that CMS.
发布评论
评论（4）
对于扩展 CMS 站点，通常需要与第三方工具兼容。
我最近的经验是使用 Drupal，但这些工具与其他基于 *AMP 的工具（即基于 PHP/Apache/MySQL）是相同的，就像大多数开源 CMS 一样。
您需要确定的真正关键问题是每个 CMS 与这些标准工具的兼容性，以实现可扩展性。这几乎比 CMS 本身的特定可扩展性功能更重要，这些功能在大多数现代 CMS 中几乎相同（积极的缓存等）。
Linux 的典型难题如下。您应该评估其中哪些部分对您很重要，哪些部分适用于您选择的 CMS。
为了结束这一切并回答您的问题，我当然推荐 Drupal 作为坚如磐石的可扩展 CMS。扩展它是否很容易，以便它运行像 examer.com 或 Whitehouse.gov 这样的网站，或者拥有数百个甚至数千个子站点和数百万个节点的大学网站？哎呀不！在您选择的任何 CMS 中，这一挑战都不容易。但如果您解决了上述所有问题，您也应该能够选择任何现代流行的 CMS 并使其快速运行。
For scaling CMS sites, it's often a case of compatibility with 3rd party tools.
My recent experience has been with Drupal, but the tools are the same for other *AMP based tools (ie. PHP/Apache/MySQL based), like most open source CMS's are.
The real key thing to determine for you will be the compatibility of each CMS with these standard tools for getting the scalability happening. That is almost more important than the specific scalability features of the CMS itself, which are pretty much the same in most modern CMS's (aggressive caching and so on).
The typical pieces of the puzzle for Linux are the following. You should evaluate which of these pieces matter to you and which will work with the CMS you choose.
And to end this all of, and to answer your question, I of course recommend Drupal as a rock solid scalable CMS. Is it easy to scale it so it runs a site like examiner.com or whitehouse.gov or a university site with hundreds if not thousands of sub sites and millions of nodes? Heck no! Nor will that challenge be easy in any CMS you choose.. but you should also be able to choose any modern popular CMS and make it work fast, if you address all of the above.
Plone 有一个非常好的负载平衡故事：
http://developer.plone.org/reference_manuals/active/部署/
http://developer.plone.org/reference_manuals/active/deployment/stack.html#load-balancing
Plone has a very good loadbalancing story:
http://developer.plone.org/reference_manuals/active/deployment/
http://developer.plone.org/reference_manuals/active/deployment/stack.html#load-balancing
我们目前在负载相当高的集群设置中使用 Alfresco。工作正常。不过，我们注意到 lucene 索引同步存在一些小问题 - 它可能需要一两分钟才能通过所有节点进行复制。
GA 版本是开源的。他们还提供广泛的文档和培训。
We are currently using Alfresco in a clustered setup with rather high loads. Works fine. We have noticed minor issues with lucene index synchonization though- it may take a minute or two for it to replicate through all nodes.
The GA version is open source. They also provide extensive documentation and training.
这是我的观点，有些人可能不同意，但大多数中型企业内容管理系统（开源和商业）都提供负载平衡功能。问题是他们的解决方案的效率和可扩展性如何。例如，Sitefinity 提供负载平衡，其中网络场中的每台服务器都必须了解其他服务器。例如，如果我的网络场中有 20 台服务器，那么我必须配置每台服务器并确保它了解其他服务器。这确实很痛苦，但它仍然是一个“负载平衡”功能。我想说的是，在决定使用某个 CMS 之前，请确保您调查了负载平衡在该 CMS 中的工作原理。
This is my opinion and some might not agree with it, but the majority of medium to enterprise content management systems (both open source and commercial) offer load balancing functionality. The question is how efficient and scalable is their solution. For example, Sitefinity offers load balancing where each server in a web farm must be aware of other servers. For example, if I have 20 servers in a web farm, then I have to configure each server and make sure it's aware of other servers. This is a real pain, but it's still a "load balancing" feature. What I'm trying to say is make sure that you investigate on how the load balancing works in a certain CMS before you decide to use that CMS.