|Wordpress : Using accelerators better than plug-in's|
We have 20,000 page WP site which had W3 Total Cache web cache plugins installed by our developers. Our developers are not server savvy, by their own admission and have not done sites in excess of 100 pages. But I did not think this would be a big deal. Not long after the web cache the server crept to a halt ( 500 server error ). Not sure if it was connected but over 80,000 files were held in the cache folder, which our administrator ditched.
Being concerned, I asked our server administrators what to do, and this is what they said :
"There are alternatives to cache for wordpress. I was researching that this can actually work against the purpose. There are solutions such as varnish cache, which are installed at a server level (meaning we would need to install and configure this, versus the developer ) which can take the server load from maxing out with 15 users on wordpress, to allowing 1000s of users with no effect on performance. Part of the problem appears to be wordpress, it doesn't perform as well as it used to."
Our sites do operate IP look up's to display the language and currency of user - so not sure if any of these server side implementations will upset this.
A while back during a period of instability on an old server another administrator said that we should install NGIX and Varnish. he also asked at that stage that all plugins for forms and emails be disengaged, until we moved to a more robust server configuration. Which we now have.
I thought Wordpress can handle substantial scale in terms of traffic and pages.
Do any of the experts that frequent these threads have a view of how to manage a stable implementation with regards to Wordpress and speeding up larger sites, in simple language?
WordPress can handle what you have. I'm pretty sure WordPress.com is a single multisite installation with over 35 million subsites. The key is in the setup.
WordPress.com uses a cloud based hosting environment but that's about as much as I know of the setup. I'm sure you can find out - even ask Automattic staff.
Confirmed that WordPress.com IS one huge multisite install.
Have you looked at this WordPress resource?
Seems like the popular " w3c total cache " plug-in , demonstrates an example of what can happen on problems, per the OP.
In this case, there is a limited number of inodes on a server, and w3c total cache had used up the total amount allowed for the server, and as such gave the same error as the disk drive being full and provided a 500 error. Deleting the cache reduced the number of used inodes on the server.
On the threads I did notice great disruption out there and scrambling to find alternatives, hosting companies threatening closure of accounts.
Back to the OP - maintenance / updates / accountability for plugin's ? Can you trust plug-ins v more widely used alternatives to speed such as accelerators ?
W3C Total is an excellent plugin. It does take some knowledge to configure it AND maintain but the results are well worth it.
It has so much depth, that it is difficult for most people to get it right.
If you do not have the experience, I would suggest you hire the developers to install and fine-tune it for you, and perhaps make suggestions for a better host if yours is not up to scratch.
Are you leveraging the visitors' browsers caching resources via Etags or other caching options? WP does not set these by default. Caching plugins and accelerators (pre-fetch)use your server resources, Etags and gz settings use the visitors' browsers to speed loading of repetetive js and theme images, js and css files.
Also make sure that your images are web optimized using indexed color and lower color depth wherever possible. You can't squeeze everything out of photo images, but many background and button images are much bigger than they should be (in filesize - not dimensions). Even a little 15Kb background gif can be a 600b png file and load much faster. If repetetive images, css and js files are optimized, browser cached and deflated, your pages can load faster and your server works more efficiently.
|It does take some knowledge to configure it AND maintain but the results are well worth it. |
How much maintenance ?
I also get perplexed by the comparisons between plug-in's touted by the creators and those happy/un-happy with the experiences.
Somewhere I read, for example that Super Cache is far simpler to configure than w3c total cache ; then I saw 3.5m downloads v 2.5m downloads. It's hard to fathom. As a business owner, I just want to know our sites are going to scale and provide minimal overhead.
I take the advice above about getting the plugin creators involved in configurations, but having different parties sharing in changes under the hood, is a business owner's nightmare if something goes wrong. In particular if one of them just doesn't agree on something or a problem remains unsolved clamed on another party, etc.
So in the context of this, isn't it better to preference an accelerator over a plug-in.
We've used both for years, and yes SuperCache is very simple and easy to install for most anyone. Having far less capabilities is what makes it easy and I guess most people will prefer this.
W3C Total cache as I said has lots of depth but you do not have to use everything. Using just the defaults will get you the same results as SuperCache.
It is when you start to delve into the deeper features is when it blows past SuperCache. The more deeper you get the faster it gets but the more delicate it is to fine tune. So you need to be an expert or use the developers. The Pro version interfaces with a specific WP framework and opens up even more speed.
With both plugins, it helps make life simpler to cut down the fat in the number of plugins you run as some of these may conflict with the accelerator.
If you are not keen on running accelerators the option is to lease a very fast dedicated server from companies that specialize in WP hosting.
So using any form of cache appears to be about the configuration. Knowing what to exclude from cache and what to include as well.
It seems to me that the business model of these cache's is driven around the complexity which causes site owners to require the inputs of the creators. I guess nothing is ever for free.
There seem to be alternatives to Varnish such as [cloudflare.com...]
It looks like it's free. I've been told it creates a layer in-between the server and the users. This increases security and adds a proxy on static content the same way cache works. I don't know anyone who uses it and the one thing I am not 100% sure of is if it's necessary to sign over the DNS management of the domain to them to look after. Which is a bit of a pain as all DNS are under a wholesale registry panel for us.
In our case we have multiple domains, which operate on the same templates and the same caching principles [ I assume ] across all sites.
Whitey I've looked at some of your other posts and it appears your sites have thousands of pages.
If you are also fielding hundreds of simultaneous visitors, don't you have the budget for fast and dedicated servers run by WP professionals?
That would take away all the headaches and minutia of dealing with servers and plugins and allow you to write and develop those thousands of pages.
Really, dealing with accelerators can drive you batty ;o)
@romero - these sites are part of a new batch, built on WP which is relatively unfamiliar to me. Other sites I've been associated with had bespoke development and server admin support tailored to their needs. The objective is to get some sites positioned driving a modest level of organic traffic before reviewing the platform and setup for the next level of scale. So budget is restricted for these and x no of sites is an expense I'm wary of versus real value.
Because of the restricted budgets, we've engaged WP developers who have only handled sites with around 1-200 pages max, but churn out good sites at the rate of around 40-50 per month. Server admin looks after a large number of enterprise level sites as well as small sites. But as always diversified handling strains the knowledge graph.
I'm kinda starting to downgrade the significance of the plugins / accelerators for factors effecting SEO performance as server response time has been dropped from 1.48secs to 0.48secs recently, without plugins or accelerators, due to switching of the dysfunctional plugin and file cleaning. A target of .20 secs would be where I'd like it as part of an overall target score of 90/100 on the Google site insights scale for desktop and mobile.
I guess you see speed as important , ( and others with 5-6 million downloads between the two major cache plugin creators ) since you do it. But how important at this level?