the size of the databases ahave nothing to do with the RAM.
The essence of using databases is that they occupy more space than we have available RAM memory.
The RAM is used as cache, so for exemple if you have 10 Databases with each uses 20MB of cache you have 10*20MB=200MB of RAM Cache used.
All my bases are: one 170KB and the other one: 240MB. And the essence of using the cache is "availability of the most used fragments of the database"
And clearly "269074432" == +/-256MB
I see that at least in the calculation we agree.
So you seem to saturate your private SQL.
I moved back to ISAM, the situation looks better, the bases use about 120MB of available 256MB. However, from time to time the database engine resets. It is not possible to add a new (simple) index to the largest of the tables (120MB), the database engine also resets with the given message (and no one else at the time is not working, web users are blocked)
I think the configuration of a private SQL server is badly done.
When I worked (and with larger databases) on a previous version of a private SQL server (with 128MB of RAM) then this was not the case, the server operation slowed down (understandable) but the reset did not occur.