DISQUS

Using a DB to mimic could be faster than just using the filesystem in some instances. You don't have to open() each file for each access, assuming an appropriate level of caching. Also, retrieving multiple entries could in theory be pulled from disk in only one read as opposed to multiple.

Also as mentioned below, you don't get indexing if you just use the filesystem.

The biggest problem is that this isn't strictly a relational / non-relational problem. It's a mix, and so this could be an appropriate solution.
I might have gone for a different solution, but it seems to work for them.

Given the history of "every other cell phone manufacturer and every other cell phone carrier" when it comes to producing phones with decent interfaces, I think the iPhone looks pretty safe.

I wouldn't count out Google in building a good platform, but I doubt that the carriers could help themselves to tweak (lock-down) things here and there in closed-sourced ways.

The real problem is that Hollywood demanded the ability to restrict the computer hardware that you use. And Microsoft took their side. So now I can't use my high res LCD monitor to watch movies from Netflix, but I can use my crappy old CRT? I don't blame Netflix. I blame Microsoft.

How is this progress?

Actually, you need 4 bits to store nucleic acid sequences, if you include ambiguity codes as well... (http://www.bioinformatics.org/SMS/iupac.html)

0001 A
0010 C
0100 G
1000 T

Now, if you don't know the base, it's 1111 (N), if it's an A or a C, it's 0011, etc...

Using bases as a bitmap also makes comparisons much faster too... you can just AND each bitmap against each other and if the result is greater than zero, it's a match.

There is a difference between MySQL Cluster and MySQL replication. What you seem to be describing is replication, in which case I don't think that the linked article will help you.

It seems to me that your approach should work out just fine. Just don't try to get into master-master replication... it's a pain and it can only lead to tears. :)

If you want to make it more robust, you could add a couple of features to your daemon. For example, you should pull your config information from a file. Then you could implement a simple protocol. For example, the client could send "REQ" to request a DB handle. Or the client to send "RELOAD" to cause the daemon to reload it's config file. This way you wouldn't have to stop the daemon...

Then again, PHP probably isn't the best language to use for the daemon, since it will block on each request. If you have two requests coming in at the same time, the second one will have to wait for the first to get done. This might be an issue over time.