Memcached or MySQL for session storage - PHP - mysql

Is it good idea to use Memcached for session storage with PHP? We will have a lot of servers and we must access the session data from everywhere so we are forced to use database (in our case that will be MySQL) as session storage or Memcached. What do you think?

I know people who've used Memcached for this -- it's very fast, certainly a lot faster than a database, and is built to handle a lot more concurrency.
The primary disadvantage to purely in-memory storage is that all your session data will be wiped if/when you restart the daemon. In my experience, memcached is rock-solid and I've never had to restart it because of a failure, but it is a consideration if your sysadmins aren't used to working that way, or if your systems are updated frequently. It also depends on whether losing all your user sessions once a month or year is acceptable or not (i.e. in ecommerce, management probably won't like this).
The obvious solution, if that's the case, is to go to one of the many disk-based NoSQL/hash table databases, such as MemcacheDB, which is based off of Memcached. Or see: CouchDB, MongoDB etc. Each of these daemons (including Memcached) is also a lot less complex when it comes to performance tuning than MySQL (where all sorts of things like key and sort buffers, query cache etc. have to be tuned per install/use case) -- I mean, with Memcached there's not much to do other than to allocate memory and start it up.
Personally, I am a fan of using faster, more appropriate (non-SQL) storage for temporary things like session keys, but if your database is not under load and you don't anticipate it to be, the only thing you lose by storing sessions in the database is that it's a little slower, so users see a little more latency.
Whichever way you go, I suggest that you write your session-management code in such a way that the storage engine is just a layer, and you can swap in a different storage engine relatively painlessly. You don't want to be recoding your application if you find memcached or whatever you choose isn't working well, and you want to try something else. For instance, I once wrote a caching system for a clustered CMS application that used memcached to cache various pages and objects, but when the daemon wasn't reachable, it would fail over to alternate backends that would cache to shared memory or disk on the individual webservers. (In your case, you don't necessarily need the auto-failover, just the ability to change your mind about the backend.)
I mentioned MemcacheDB because it uses the Memcache protocol, so it's extremely easy to swap in Memcached for MemcacheDB or vice versa.

Related

Should I implement my own caching or rely on read-replicas?

We have an enterprise application that uses an SQL database. The database access characteristics are about 90% reads. The data that does get updated or created needs to be up-to-date immediately. The cache needs to be correctly invalidated with high certainty. The entities are referred to by their primary key for 98% of the cases.
The application is based on Node.js and is AWS-native. Since the application is AWS-native, I'd like to rely on managed services from AWS rather than hosting my own. One option is to implement our read-through Redis-based cache. Upon retrieving the entities, we'd check the cache and if the data is not cached we'd put it into the cache before turning it to the user. The parts of the code that update those entities will invalidate the cache by primary key.
Generally speaking, in computer science cache coherency is one of the most challenging problems to get right. I am of the opinion that rather than implementing a Redis cache and thinking through all of the possible scenarios for correctly invalidating it, it is wiser to instead configure an Aurora read-replica specifically for reading frequently accessed entities. The RDBMS will do a much better job at caching than anything we can build ourselves.
So, I am facing two options -- go through the effort of implementing my own caching, or use read replicas. My personal opinion is to use a read replica.
Any advice is greatly appreciated, as always.
Yes, you're right, cache invalidation is a tough problem. The simplest solution is to add code to your data writes, to replace the cached values. So they're always current. But this is easy only if the cached values have a pretty much 1-to-1 correlation with rows in your database.
An advantage of your own cache is that you can cache data that is not 1-to-1 with rows of data in the database. You might cache an entire HTML fragment for a drop-down menu for example. That could be the result of several SQL queries. It could be quite an advantage to cache data that is higher up the "food chain" so to speak. But cache invalidation becomes less straightforward. Best for storing results of queries that don't change often.
Using a read-replica is not a substitute for using a cache. Querying a read-replica still has overhead of making a database connection, authentication, SQL query parsing and optimization, locking, and all the other overhead that goes into RDBMS workings.
Querying data from a cache can be orders of magnitude faster.
Both have their place. It's best to use both a cache and a read-replica for different tasks. I would also add message queues as an important technology. I believe database, cache, and queue form a three-legged stool.
But you must have experience and judgment to know when each is the best tool for a given case.

handling stale caches with multiple servers

We currently have a website hosted on one server, and we are looking into adding a new server. The main issue is about caching. Some items are cached based on when they are changed. However right now, they are changed in the same process, hence the cache can be invalidated.
If the website is hosted on two servers, the changes can be done on both servers and they will not be notified of such changes. The cache needs to remain as it drastically speeds up the website. I would prefer if the cache is not taken out-of-process in a cache-server, as it slows down to the speed of network rather than memory, and adds complexity to the servers.
The website is implemented in .Net, with MySQL as it's backing datastore. My issue is how the process can be notified when data changes. Is it possible that MySQL will automatically notify all registered clients when any data changes? I've used RavenDb, which has a similar feature which comes in very handy. I couldn't find anything similar for MySQL. If this is not possible, any ideas how one would approach this issue?
Distributed caching is a complex topic. It sounds like you are running a more basic in-memory cache. If this is the case, you will need to handle synchronisation yourself, or be happy with "eventual consistency" of the data, assuming you have some stale key checking mechanism.
Personally I would look into using memcached (we use Couchbase). Your opinion on this becoming a network bottleneck may be unrealised, although yes in real terms memory access is faster. In practical terms, we noticed that Couchbase caching was more than fast enough, and it is atomic at the key level. It will handle key distribution over nodes.
As for MySQL pushing notifications to clients, I am not sure but I don't think so. You could emulate this yourself if you have a layer of code (DAL etc) over database access.
It is also difficult to reconcile the desire to have the cache follow the same integrity principles as the database. If you achieve this then all you have done is made an in-memory database. Caching is supposed to be a trade-off of data accuracy over time to increase scalability.

Seeking clarification about mysql 5.6 memcache integration

I'm having trouble getting a clear understanding of what MySQL 5.6 is introducing w/r/t memcache.
As I understand it, memcache by itself is essentially a huge, shared, memory-resident hash table that is managed by a server, memcached. In particular, it knows nothing about a persistent data store, and offers no services in that regard. It simply knows about keys and values (like a Perl hash).
What I think mySQL 5.6 introduces is a NoSQL API, whereby mySQL clients can request data from the mySQL server by key, rather than by a SELECT statement. (And similarly, they can perform updates with key=value pairs). MySQL uses memcached to cache these in memory as a performance boost, but also takes care of things like writing updates back to the database before they age out of the cache, etc.
In other words, the use of memcached is an implementation detail of the mySQL 5.6 NoSQL feature, and is not something the application programmer needs to be aware of.
I'd welcome any corrections or amplification to my understanding.
Thanks,
Chap
I think it's quite simple (from the official documentation):
I disagree with your last sentence, the application programmer has to be really aware of the memcache plugin because having it onboard of the MySQL server means that he can decide (maybe he will be forced to) access data through a memcached language interface or via the SQL interface
To better understand the impact of this plugin onto an app design you should know that there are 3 configuration tables used by MySQL for a proper memcached management; understanding how the "cache_policies" works will shade some light to some of your doubts:
Table cache_policies specifies whether to use InnoDB as the data store of memcached (innodb_only), or to use the traditional memcached engine as the backstore (cache-only), or both (caching). In the last case, if memcached cannot find a key in memory, it searches for the value in an InnoDB table.
here is the link: innodb-memcached-internals
This quote above means that, depending on what you decided for a specific key-value, you will have different application scenarios :
innodb_only -> means that you can query the data via a sql interface or via a memcached interface, here is a link to some memcached language interface examples memcached-interfaces
cache-only -> means that you should query the data via the memchached interface only
caching -> means that you can use both the interfaces (note that the storage mechanism slightly changes)
Of course this latter configuration decision is strictly related to your specific needs
I don't really have a complete answer for you I'm afraid, as I too am struggling to find the detail I require before toying around with it.
That said however there is one important point which I have managed to uncover that you seem to have missed, namely that by accessing the InnoDB storage engine via the new plugin you are actually completely bypassing SQL and avoiding all the overhead that comes with it.
This of course makes it essentially a key/value store more akin to most NoSQL databases complete with all the drawbacks associated with them. i.e. no joins etc...
However on the flip side for many applications these days, this is exactly what we want. There has been only a handful of real world performance mentions that I have come across but all seem to point to this implementation significantly outperforming MongoDB and other similar NoSQL solutions (how much truth is in it I do not know) with even one (relatively in depth) comparison claiming as high as 700k qps on a commodity server (compared with around 100k on a well tuned MySQL setup), which is incredible if true.
Resource here:
http://yoshinorimatsunobu.blogspot.co.uk/search/label/handlersocket
Anyway, sorry I can't be any more help but its food for thought at least!

Which database for a web crawler, and how do I use MySQL in a distributed environment?

Which database engine should I use for a web crawler, InnoDB or MYiSAM? I have two PC's, each with 1TB hard drives. If one fills up, I'd like for it to save to the other PC automatically, but reads should go to the correct PC; how do I do that?
As for the first part of your question, it rather depends on you precise implementation. If you are going to have a single crawler limited by network bandwidth, then MYiSAM can be quicker. If you are using multiple crawlers then InnoDB will give you advantages such as transactions which may help.
AFAIK MySQL doesn't support the hardware configuration you are suggesting. If you need large storage you may wan tot look at MySQL Cluster.
MyISAM is the first choice, because you will have write only operations and crawlers -- even run in parallel -- will be configured -- I suppose -- to crawl different domains/urls. So you do not need to take care of access conflicts.
When writing a lot of data, especially text!, to Mysql avoid transactions, indexes, etc., because it will slow down MySQL drastically.

Best storage engine for constantly changing data

I currently have an application that is using 130 MySQL table all with MyISAM storage engine. Every table has multiple queries every second including select/insert/update/delete queries so the data and the indexes are constantly changing.
The problem I am facing is that the hard drive is unable to cope, with waiting times up to 6+ seconds for I/O access with so many read/writes being done by MySQL.
I was thinking of changing to just 1 table and making it memory based. I've never used a memory table for something with so many queries though, so I am wondering if anyone can give me any feedback on whether it would be the right thing to do?
One possibility is that there may be other issues causing performance problems - 6 seconds seems excessive for CRUD operations, even on a complex database. Bear in mind that (back in the day) ArsDigita could handle 30 hits per second on a two-way Sun Ultra 2 (IIRC) with fairly modest disk configuration. A modern low-mid range server with a sensible disk layout and appropriate tuning should be able to cope with quite a substantial workload.
Are you missing an index? - check the query plans of the slow queries for table scans where they shouldn't be.
What is the disk layout on the server? - do you need to upgrade your hardware or fix some disk configuration issues (e.g. not enough disks, logs on the same volume as data).
As the other poster suggests, you might want to use InnoDB on the heavily written tables.
Check the setup for memory usage on the database server. You may want to configure more cache.
Edit: Database logs should live on quiet disks of their own. They use a sequential access pattern with many small sequential writes. Where they share disks with a random access work load like data files the random disk access creates a big system performance bottleneck on the logs. Note that this is write traffic that needs to be completed (i.e. written to physical disk), so caching does not help with this.
I've now changed to a MEMORY table and everything is much better. In fact I now have extra spare resources on the server allowing for further expansion of operations.
Is there a specific reason you aren't using innodb? It may yield better performance due to caching and a different concurrency model. It likely will require more tuning, but may yield much better results.
should-you-move-from-myisam-to-innodb
I think that that your database structure is very wrong and needs to be optimised, has nothing to do with the storage