I’m no fan of PHP or Wordpress by any stretch. But I agree with the pragmatic choice in this case. It will probably only need one dev to maintain the site. But even that would be overkill. You just never want a bus factor of 1. Especially for such a high profile site. But I’m digressing...
But cloud vs VPS aside, I do wonder is they are self hosting due to security reasons. Imagine a bad actor hacking into the WP installation! It does have a bad security record.
Stuff like NFS made sense for on prem but it’s a terrible solution for building services on AWS (for example).
Security wise, any popular CMS will be a security nightmare. It’s not even a PHP thing. In an ideal world you’d stick your CMS behind a WAF and put some additional policies around your admin APIs (if at all possible). Wordpress, for all of its warts, actually made those parts easy to do.
I disagree, and I think it comes from a different understanding of what web security is.
Popular CMS software benefits from millions of people hammering on installations. If the community constructs a sane process for handling vulnerability reports (and WP has), then the average user of that software benefits from a robust continuous testing regime that they could never afford themselves. Testing and patching is an extremely important component of web application security in the long run, and also very expensive to do well. Most custom CMS projects do not bother to do it at all, and yet fool themselves into thinking they are secure because they wrote their own code.
No one can write secure software on the first try. The most well-funded and well-regarded software companies in the world spend tremendous effort on coding practices and testing, and still ship vulnerabilities.
Continuous testing and patching is always required. If your website is the only one running your CMS software, your first hint of a vulnerability may be when it gets exploited in production.
Whenever a new vulnerability is discovered for WP (or any popular suite, be it a shop like Magento or message board like phpBB) you then get hordes of bots that trawl the internet looking for sites that run out of date versions of said software just for the purpose of hacking it. Even search engines become tools for bots to hunt out these insecure platforms.
I’m not someone who advocates security through obscurity however the vast majority of WP et al hacks are not targeted and thus a home grown CMS technically wouldn’t fall foul in the same way.
HOWEVER please don’t twist my words into saying that home grown CMSs are “more” secure. I’m making no such claim what-so-ever (and nor did I in my previous post when you want off on a soliloquy about how I apparently suggested that — I did no such thing!). My statement is and was only that any popular CMS will be a security nightmare. That point is very much true.
This is why I also brought up the point about WAF and additional policies on your admin path. The former buys you a little more time with updates (assuming you’re with a reputable WAF that supports the CMS platform you’re hiding against) because they can identify suspicious requests (ie ones that look like they’re triggering a known vulnerability) and simply block those requests from your site (technically the WAF is a reverse proxy with some firewalling, request heuristics, etc).
As for the point about admin policies: if you can IP whitelist who gets access to /admin then that saves you a lot of pain early on. But that’s obviously not always practical. However there are plenty of other ways to harden access.
Disclaimer: in a former life I used to repair and harden hacked CMSs, shops like Magento, etc.
We always aimed to evaluate security patches same-day. The shortest lag we saw was “Drupalgeddon”—about 12 hours. In some cases we would see a lag of up to a month between the release of a patch and the signatures of automated attempts to exploit the underlying vulnerability. We eventually got rid of our IDS because all it did was log failed attempts to exploit vulnerabilities we had already patched, and fruitless door-knocking like password brute force attempts.
I’m sure you saw a lot of CMS installations that got hacked. I’m also sure that most them were because of either poor patch discipline, or misconfiguration, or both.
I guess it’s true that if someone is going to be bad at running a website, a popular CMS might be a bad experience for them. But even then a popular CMS may have an advantage: it’s a lot easier to find professional help for a popular CMS than for a niche or custom one.
Honestly, running any software at all is going be a nightmare for a lot of people. Hence the success of products like Squarespace and Wix; now they don’t have to.
All that means is that you’ve misconfigured your IDS. You’re supposed to filter out stuff that’s not a threat, not remove the entire IDS instance entirely!
You shouldn’t throw the baby out with the bath water https://en.m.wikipedia.org/wiki/Don't_throw_the_baby_out_wit...
Again you’re over estimating the average IT team’s ability to keep on top of WP patches. And again I have to reiterate my point about the advantages of a good WAF (web application firewall).
I’m not making a theoretical argument here. I’ve seen and have had to fix numerous sites from companies with competent developers and sys admins but who were simply unlucky enough not to notice that one RCE vulnerability in time.
> I’m sure you saw a lot of CMS installations that got hacked. I’m also sure that most them were because of either poor patch discipline, or misconfiguration, or both.
Neither actually. Just busy teams in medium sized business having a plethora of projects and too little time to manage everything with the same degree of love that your larger companies or start ups do when they have a dedicated team working in a single product.
Remember the appeal of WP for many is that it is “plug and play” so it often gets installed in organisations as a way of reducing the burden on existing developers and/or sysadmins. A strategy that makes complete sense on paper but can back fire spectacularly.
Let’s also remember that WP also appeals to the hobbyists too. You can’t accuse them of being incompetent sysadmins because they never set out to be technical. Instead they were sold on how easy WP is to install and manage.
Again this is why I keep discussing WAFs. Using them is non-obvious but it does reduce the burden on whatever individual or team manage WP.
> But even then a popular CMS may have an advantage: it’s a lot easier to find professional help for a popular CMS than for a niche or custom one.
Oh i agree. Choosing an established CMS isn’t a bad decision as there are plenty of benefits in doing so.
> Honestly, running any software at all is going be a nightmare for a lot of people. Hence the success of products like Squarespace and Wix; now they don’t have to.
This I agree with too.
I’ve seen WP used for everything from personal blogs, international news sites, shopping sites, and even fantasy sports games. The diversity of WP is definitely a strength. But it does mean that there’s not going to be a 1 size fits all alternative.
There's no "NFS requirement for multiple Wordpress hosts" tho.
Unless you have a very dynamic site (posts hourly, multi-daily), would making it static-y be a solution?
Do a post via web GUI, it gets written out to disk, and then you rsync (or auto-git or whatever) to distribute it.
However if you can and are going down the static site route then I’d wager there are easier platforms to start with.
I built a static site generator that takes markdown files and “compiles” them to HTML using pandoc. The whole thing is just 1 shell script. But that use case also wouldn’t suite most other people.
So it really depends on what the team need and what the site does.
Often the “loads of ways” on WP require changing its normal behaviour and thus breaks ones ability to use the admin console. Or use some other kludge that might suit some subset of people some of the time. But ultimately it keeps coming back to kludges or other workarounds that introduce as many caveats as they solve.
Wordpress made sense when the web was a collection of independent servers. However in today’s cloud-orientated hosting landscape there are a thousand better ways to provide said content. Both at the beginner level and at the enterprise.