I have an app...
The app does a market comparison for a financial product - for a given quote request, it contacts several other sites for their quotes. It then gives the user the results - several quotes for their details.
To manage these requests they get saved to MySQL and then my app kicks in, picking up the pending quotes and farms these out to threads (all same Linux box) to process each site lookup.
I am using JRuby as I had thread/db related issues. Using Java threadpools to control the number of threads. With the current hardware/VPS - it can handle around 200 threads. A lot of the limitations seem to relate to each thread grabbing their own MySQL connection - grabbing the quote details and saving back the results. We want to handle more concurrent threads and so looking for ways to scale up.
Wondering which way to go ...
Bigger hardware...
More machines and use some kind of queueing
mechanism (with priorities) to share the load across the machines -
so the threads dont touch the db, all the details/responses go via
the queue - so the DB hit is less, but then maybe I am just pushing
the problem into the queue. Thinking of using something like
MongoDB for the queue, but open to suggestions - something easy to
use with Ruby :)
Some kind of remote/RPC mechanism, eg dRb -
theoretically this seems like a good option, but not done anything
with this yet to know how complex it will make things.
Something
else...?
From this link Reasons for NOT scaling-up vs. -out? - it would seem this problem is suited to running more machines to solve it.
So, any thoughts on which way to go...
Cheers,
Chris
My usual approach to problems like this is to pay very close attention to the database queries you're making and tune them aggressively. Retrieve only what you need, skipping columns that aren't explicitly used, and be very careful about eager loading things you don't need in their entirety.
You'll often find you can get significant speed gains by adding indexes, or strategically de-normalizing certain attributes in your database to avoid ugly, time-consuming JOIN operations.
Further, think about caching: The fastest database call is the one that's never made. It's not hard to leverage in something like Memcached to save the results of a moderately time-consuming record retrieval and if done carefully it's even easy to invalidate and expire this provided you channel your updates through a few methods.
For scheduling workers, a simple first-in, first-out queue can be implemented in Redis to off-load a lot of the processing overhead from MySQL itself. This is usually very simple to add if you follow an example.
A cache like Memcached can handle an extremely high amount of traffic, so whenever possible, cache against this to avoid hitting your database for every last thing.
If you've exhausted these options, it's time for more front-end servers and even more database capacity, but only then.
Queing is easiest thing for you to implement. Use something like this: http://beanstalkd.github.com/beaneater/
Basically you can prepend your methods with async. which will put them into queue and execute them. They queue and workers can be same server or a different one.
We are at the beginning of a new project, and we are really wondering if we should use stored procedures in MySQL or not.
We would use the stored procedures only to insert and update business model entities. There are several tables which represent a model entity, and we would abstract it in those stored procedures insert/update.
On the other hand, we can call insert and update from the Model layer but not in MySQL but in PHP.
In your experience, Which is the best option? advantages and disadvantages of both approaches. Which is the fastest one in terms of high performance?
PS: It is is a web project with mostly read and high performance is the most important requisite.
Unlike actual programming language code, they:
not portable (every db has its own version of PL/SQL. Sometimes different versions of the same database are incompatible - I've seen it!)
not easily testable - you need a real (dev) database instance to test them and thus unit testing their code as part of a build is virtually impossible
not easily updatable/releasable - you must drop/create them, ie modify the production db to release them
do not have library support (why write code when someone else has)
are not easily integratable with other technologies (try calling a web service from them)
they use a language about as primitive as Fortran and thus are inelegant and laborious to get useful coding done, so it is difficult to express business logic, even though typically that is what their primary purpose is
do not offer debugging/tracing/message-logging etc (some dbs may support this - I haven't seen it though)
lack a decent IDE to help with syntax and linking to other existing procedures (eg like Eclipse does for java)
people skilled in coding them are rarer and more expensive than app coders
their "high performance" is a myth, because they execute on the database server they usually increase the db server load, so using them will usually reduce your maximum transaction throughput
inability to efficiently share constants (normally solved by creating a table and questing it from within your procedure - very inefficient)
etc.
If you have a very database-specific action (eg an in-transaction action to maintain db integrity), or keep your procedures very atomic and simple, perhaps you might consider them.
Caution is advised when specifying "high performance" up front. It often leads to poor choices at the expense of good design and it will bite you much sooner than you think.
Use stored procedures at your own peril (from someone who's been there and never wants to go back). My recommendation is to avoid them like the plague.
Unlike programming code, they:
render SQL injection attacks almost
impossible (unless you are are
constructing and executing dynamic
SQL from within your procedures)
require far less data to be sent over
the IPC as part of the callout
enable the database to far better
cache plans and result sets (this is
admittedly not so effective with
MySQL due to its internal caching
structures)
are easily testable in isolation
(i.e. not as part of JUnit tests)
are portable in the sense that they
allow you to use db-specific
features, abstracted away behind a
procedure name (in code you are stuck
with generic SQL-type stuff)
are almost never slower than SQL
called from code
but, as Bohemian says, there are plenty of cons as well (this is just by way of offering another perspectve). You'll have to perhaps benchmark before you decide what's best for you.
As for performances, they have the potential to be really performant in a future MySQL version (under SQL Server or Oracle, they are a real treat!). Yet, for all the rest... They totally blow up competition. I'll summarize:
Security: You can give your app the EXECUTE right only, everything is fine. Your SP will insert update select ..., with no possible leak of any sort. It means global control over your model, and an enforced data security.
Security 2: I know it's rare, but sometimes php code leaks out from the server (i.e. becomes visible to public). If it includes your queries, possible attackers know your model. This is pretty odd but I wanted to signal it anyway
Task force: yes, creating efficient SQL SPs requires some specific resources, sometimes more expensive. But if you think you don't need these resources just because you're integrating your queries in your client... you're going to have serious problems. I'd mention the analogy of web development: it's good to separate the view from the rest because your designer can work on their own technology while the programmers can focus on programming the business layer.
Encapsulating business layer: using stored procedures totally isolates the business where it belongs: the damn database.
Quickly testable: one command line under your shell and your code is tested.
Independence from the client technology: if tomorrow you'd like to switch from php to something else, no problem. Ok, just storing these SQL in a separate file would do the trick too, that's right. Also, good point in the comments about if you decide to switch sql engines, you'd have a lot of work to do. You have to have a good reason to do that anyway, because for big projects and big companies, that rarely happens (due to the cost and HR management mostly)
Enforcing agile 3+-tier developments: if your database is not on the same server than your client code, you may have different servers but only one for the database. In that case, you don't have to upgrade any of your php servers when you need to change the SQL related code.
Ok, I think that's the most important thing I had to say on the subject. I developed in both spirits (SP vs client) and I really, really love the SP style one. I just wished Mysql had a real IDE for them because right now it's kind of a pain in the ass limited.
Stored procedures are good to use because they keep your queries organized and allow you to perform a batch at once. Stored procedures are normally quick in execution because they are pre-compiled, unlike queries that are compiled on every run. This has significant impact in situations where database is on a remote server; if queries are in a PHP script, there are multiple communication between the application and the database server - the query is send, executed, and result thrown back. However, if using stored procedures, it only need to send a small CALL statement instead of big, complicated queries.
It might take a while to adapt to programming a stored procedure because they have their own language and syntaxes. But once you are used to it, you'll see that your code is really clean.
In terms of performance, it might not be any significant gain if you use stored procedures or not.
I will let know my opinion, despite my toughts possibly are not directly related to the question.:
As in many issues, reply about using Stored Procedures or an application-layer driven solution relies on questions that will drive the overall effort:
What you want to get.
Are you trying to do either batch operations or on-line operations? are they completely transactional? how recurrent are those operations? how heavy is the awaited workload for the database?
What you have in order to get it.
What kind of database technology you have? What kind of infrastucture? Is your team fully trained in the database technology? Is your team better capable of building a database-aegnostic solution?
Time for get it.
No secrets about that.
Architecture.
Is your solution required to be distributed onto several locations? is your solution required to use remote communications? is your solution working on several database servers, or possibly using a cluster-based architecture?
Mainteinance.
How much is the application required to change? do you have personal specifically trained for maintain the solution?
Change Management.
Do you see your database technology will change at a short, middle, long time? do you see will be required to migrate the solution frequently?
Cost
How much will cost to implement that solution using one or another strategy?
The overall of those points will drive the answer. So you have to care each of this points when making a decision about using or not any strategy. There are cases where using of stored procedures are better than application-layer managed queries, and others when, conducting queries and using an application-layer based solution is best.
Using of stored procedures tends to be more addequate when:
Your database technology isn't provided to change at a short time.
Your database technology can handle parallelized operations, table partitions or anything else strategy for divide the workload onto several processors, memory and resources (clustering, grid).
Your database technology is fully integrated with the stored proceduce definition language, that is, support is inside the database engine.
You have a development team who aren't afraid about using a procedural language (3rd. Generation language) for getting a result.
Operations you wanna achieve are built-in or supported inside the database (Exporting to XML data, managing data integrity and coherence appropiately with triggers, scheduled operations, etc).
Portability isn't an important issue and you do not whatch a technology change at a short time into your organization, even, it is not desirable. Generally, portability is seen like a milestone by the application-driven and layered-oriented developers. From my point of view, portability isn't an issue when your application isn't required to be deployed for several platforms, less when there are no reasons for making a technology change, or the effort for migrating all the organizational data is higher than the benefit for making a change. What you can win by using an application-layer driven approach (portability) you can loose in performance and value obtained from your database (Why to spend thousands of dollars for to get a Ferrari that you'll drive no more than 60 milles/hr?).
Performance is an issue. First: In several cases, you can achieve better results by using a single stored procedure call than multiple requests for data from another application. Moreover, some characteristics you need to perform may be built-in your database and its use less expensive in terms of workload. When you use an application-layer driven solution you have to take in account the cost associated to make database connections, making calls to the database, network traffic, data wrapping (i.e., using either Java or .NET, there is an implicit cost when using JDBC/ADO.NET calls as you have to wrap your data into objects that represents the database data, so instantiation has an associated cost in terms of processing, memory, and network when data comes from and goes to outside).
Using of application-layer driven solutions tends to be more addequate when:
Portability is an important issue.
Application will be deployed onto several locations with only one or few database repositories.
Your application will use heavy business-oriented rules, that need to be agnostic of the underlying database technology.
You have in mind to do change technology providers based on market tendencies and budget.
Your database isn't fully integrated with the stored procedure language that calls to the database.
Your database capabilities are limited and your requirement goes beyond what you can achieve with your database technology.
Your application can support the penalty inherent to external calls, is more transactional-based with business-specific rules and has to abstract the database model onto a business model for the users.
Parallelizing database operations isn't important, moreover, your database has not parallelization capabilities.
You have a development team which is not well-trained onto the database technology and is better productive by using an application-driven based technology.
Hope this may help to anyone asking himself/herself what is better to use.
I would recommend you don't use stored procedures:
Their language in MySQL is very crappy
There is no way to send arrays, lists, or other types of data structure into a stored procedure
A stored procedure cannot ever change its interface; MySQL permits neither named nor optional parameters
It makes deploying new versions of your application more complicated - say you have 10x application servers and 2 databases, which do you update first?
Your developers all need to learn and understand the stored procedure language - which is very crap (as I mentioned before)
Instead, I recommend to create a layer / library and put all your queries in there
You can
Update this library and ship it on your app servers with your app
Have rich data types, such as arrays, structures etc passed around
Unit test this library, instead of the stored procedures.
On performance:
Using stored procedures will decrease the performance of your application developers, which is the main thing you care about.
It is extremely difficult to identify performance problems within a complicated stored procedure (it is much easier for plain queries)
You can submit a query batch in a single chunk over the wire (if CLIENT_MULTI_STATEMENTS flag is enabled), which means you don't get any more latency without stored procedures.
Application-side code generally scales better than database-side code
If your database is complex and not a forum type with responses, but true warehousing SP will definitely benefit. You can out all your business logic in there and not a single developer is going to care about it, they just call your SP's. I have been doing this joining over 15 tables is not fun, and you cannot explain this to a new developer.
Developers also don't have access to a DB, great! Leave that up to database designers and maintainers. If you also decide that the table structure is going to get changed, you can hide this behind your interface. n-Tier, remember??
High performance and relational DB's is not something that goes together, not even with MySQL InnoDB is slow, MyISAM should be thrown out of the window by now. If you need performance with a web-app, you need proper cache, memcache or others.
in your case, because you mentioned 'Web' I would not use stored procedures, if it was data warehouse I would definitely consider it (we use SP's for our warehouse).
Tip:
Since you mentioned Web-project, ever though about nosql sort of solution? Also, you need a fast DB, why not use PostgreSQL? (trying to advocate here...)
I used to use MySql and my understanding of sql was poor at best, I spent a fair amount of time using Sql Server, I have a clear separation of a data layer and an application layer, I currently look after a server with 0.5 terabytes.
I have felt frustrated at times not using an ORM as development is really quick with stored procedures it is much slower. I think much of our work could have been sped up by using an ORM.
When your application reaches critical mass, the ORM performance will suffer, a well written stored procedure, will give you your results faster.
As an example of performance I collect 10 different types of data in an application, then convert that to XML, which I process in the stored procedure, I have one call to the database rather than 10.
Sql is really good at dealing with sets of data, one thing that gets me frustrated is when I see someone getting data from sql in a raw form and using application code to loop over the results and format and group them, this really is bad practice.
My advice is to learn and understand sql enough and your applications will really benefit.
Lots of info here to confuse people, software development is a evolutionary. What we did 20 years ago isn't best practice now. Back in the day with classic client server you wouldnt dream of anything but SPs.
It is absolutely horses for courses, if you are a big organisation with you will use multi tier, and probably SPs but you will care little about them because a dedicated team will be sorting them out.
The opposite which is where I find myself trying to quickly knock up a web app solution, that fleshes out business requirements, it was super fast to leave the developer (remote to me) to knock up the pages and SQL queries and I define the DB structure.
However complexity is growing and without an easy way to provide APIs, I am staring to use SPs to contain the business logic. I think it is working well and sensible, I control this because I can build logic and provide a simple result set for my offshore developer to build a front end around.
Should I find my software a phenomenal success, then more separation of concerns will occur and different implementations of n teir will come about but for now SPs are perfect.
You should know all the tool sets available to you and match them is wise to start with. Unless you are building an enterprise system to start with then fast and simple is best.
I would recommend that you stay away from DB specific Stored Procedures.
I've been through a lot of projects where they suddently want to switch DB platform and the code inside a SP is usually not very portable = extra work and possible errors.
Stored Procedure development also requires the developer to have access directly to the SQL-engine, where as a normal connection can be changed by anyone in the project with code-access only.
Regarding your Model/layer/tier idea: yes, stick with that.
Website calls Business layer (BL)
BL calls Data layer (DL)
DL calls whatever storage (SQL, XML, Webservice, Sockets, Textfiles etc.)
This way you can maintain the logic level between tiers. IF and ONLY IF the DL calls seems to be very slow, you can start to fiddle around with Stored Procedures, but maintain the original none-SP code somewhere, if you suddently need to transfer the DB to a whole new platform. With all the Cloud-hosting in the business, you never know whats going to be the next DB platform...
I keep a close eye on Amazon AWS of the very same reason.
I think there is a lot of misinformation floating around about database stored queries.
I would recommend using MySQL Stored Procedures if you're doing many static queries for data manipulation. Especially if you're moving things from one table to another (i.e. moving from a live table to a historical table for whatever reason). There are drawbacks of course in that you'll have to keep a separate log of changes to them (you could in theory make a table that just holds changes to the stored procedures that the DBA's update). If you have many different applications interfacing with the database, especially if say you have a desktop program written in C# and a web program in PHP, it might be more beneficial to have some of your procedures stored in the database as they are platform independent.
This website has some interesting information on it you may find useful.
https://www.sitepoint.com/stored-procedures-mysql-php/
As always, build in a sandbox first, and test.
Try to update 100,000,000 records on a live system from a framework, and let me know how it goes. For small apps, SPs are not a must, but for large serious systems, they are a real asset.
I've been using MySQL for quite some time now. Most of that time I used it with PHP, for Joomla development. Up until now, I didn't pay very much attention to optimization, since I was usually asked to finish stuff ASAP.
Now, while I know that ASAP factor is a reality, I would like to improve my knowledge of relational DBs, together with good introspection to query and db optimization. I'm planning to start working with some rather large dbs, for which my usual approach will not be possible.
Any recommendations for some good books from the area?
Thx in advance.
Joe Celko's SQL for smarties, 4th ed.
The Art Of SQL
Refactoring SQL Applications
I would not recommend you to devote yourself to MySQL only. Instead, if you can, try to gain some experience with other DBMSs, where the advanced optimizers make your job easier.
If using linux shell I recommend mtop application to watch what is happening.
In mysql configuration you can specify logging slow queries:
http://www.webdevelopmentstuff.com/112/optimizing-mysql-log-slow-queries.html
There also is a parameter that defines what is a long query. Set it to 0 when desperate :) I once had when debugging a CMS that kept sending thousands of requests that took 0.00001s each.
I also found this: http://dev.mysql.com/tech-resources/articles/using-new-query-profiler.html
And I recommend a bit of reading on indexes.
For php+mysql with slow-queries log it's also useful to know Apache Bench command:
ab -c10 -n50 http://... calls the adress 50 times with up to 10 concurrent request.
That's just a list of tips. It's not complete in any way.
Right now I'm developing the prototype of a web application that aggregates large number of text entries from a large number of users. This data must be frequently displayed back and often updated. At the moment I store the content inside a MySQL database and use NHibernate ORM layer to interact with the DB. I've got a table defined for users, roles, submissions, tags, notifications and etc. I like this solution because it works well and my code looks nice and sane, but I'm also worried about how MySQL will perform once the size of our database reaches a significant number. I feel that it may struggle performing join operations fast enough.
This has made me think about non-relational database system such as MongoDB, CouchDB, Cassandra or Hadoop. Unfortunately I have no experience with either. I've read some good reviews on MongoDB and it looks interesting. I'm happy to spend the time and learn if one turns out to be the way to go. I'd much appreciate any one offering points or issues to consider when going with none relational dbms?
The other answers here have focused mainly on the technical aspects, but I think there are important points to be made that focus on the startup company aspect of things:
Availabililty of talent. MySQL is very common and you will probably find it easier (and more importantly, cheaper) to find developers for it, compared to the more rarified database systems. This larger developer base will also mean more tutorials, a more active support community, etc.
Ease of development. Again, because MySQL is so common, you will find it is the db of choice for a great many systems / services. This common ground may make any external integration a little easier.
You are preparing for a situation that may never exist, and is manageable if it does. Very few businesses (nevermind startups) come close to MySQL's limits, and with all due respect (and I am just guessing here); the likelihood that your startup will ever hit the sort of data throughput to cripple a properly structured, well resourced MySQL db is almost zero.
Basically, don't spend your time ( == money) worrying about which db to use, as MySQL can handle a lot of data, is well proven and well supported.
Going back to the technical side of things... Something that will have a far greater impact on the speed of your app than choice of db, is how efficiently data can be cached. An effective cache can have dramatic effects on reducing db load and speeding up the general responsivness of an app. I would spend your time investigating caching solutions and making sure you are developing your app in such a way that it can make the best use of those solutions.
FYI, my caching solution of choice is memcached.
So far no one has mentioned PostgreSQL as alternative to MySQL on the relational side. Be aware that MySQL libs are pure GPL, not LGPL. That might force you to release your code if you link to them, although maybe someone with more legal experience could tell you better the implications. On the other side, linking to a MySQL library is not the same that just connecting to the server and issue commands, you can do that with closed source.
PostreSQL is usually the best free replacement of Oracle and the BSD license should be more business friendly.
Since you prefer a non relational database, consider that the transition will be more dramatic. If you ever need to customize your database, you should also consider the license type factor.
There are three things that really have a deep impact on which one is your best database choice and you do not mention:
The size of your data or if you need to store files within your database.
A huge number of reads and very few (even restricted) writes. In that case more than a database you need a directory such as LDAP
The importance of of data distribution and/or replication. Most relational databases can be more or less well replicated, but because of their concept/design do not handle data distribution as well... but will you handle as much data that does not fit into one server or have access rights that needs special separate/extra servers?
However most people will go for a non relational database just because they do not like learning SQL
What do you think is a significant amount of data? MySQL, and basically most relational database engines, can handle rather large amount of data, with proper indexes and sane database schema.
Why don't you try how MySQL behaves with bigger data amount in your setup? Make some scripts that generate realistic data to MySQL test database and and generate some load on the system and see if it is fast enough.
Only when it is not fast enough, first start considering optimizing the database and changing to different database engine.
Be careful with NHibernate, it is easy to make a solution that is nice and easy to code with, but has bad performance with large amount of data. For example whether to use lazy or eager fetching with associations should be carefully considered. I don't mean that you shouldn't use NHibernate, but make sure that you understand how NHibernate works, for example what "n + 1 selects" -problem means.
Measure, don't assume.
Relational databases and NoSQL databases can both scale enormously, if the application is written right in each case, and if the system it runs on is properly tuned.
So, if you have a use case for NoSQL, code to it. Or, if you're more comfortable with relational, code to that. Then, measure how well it performs and how it scales, and if it's OK, go with it, if not, analyse why.
Only once you understand your performance problem should you go searching for exotic technology, unless you're comfortable with that technology or want to try it for some other reason.
I'd suggest you try out each db and pick the one that makes it easiest to develop your application. Go to http://try.mongodb.org to try MongoDB with a simple tutorial. Don't worry as much about speed since at the beginning developer time is more valuable than the CPU time.
I know that many MongoDB users have been able to ditch their ORM and their caching layer. Mongo's data model is much closer to the objects you work with than relational tables, so you can usually just directly store your objects as-is, even if they contain lists of nested objects, such as a blog post with comments. Also, because mongo is fast enough for most sites as-is, you can avoid dealing the complexities of caching and generally deliver a more real-time site. For example, Wordnik.com reported 250,000 reads/sec and 100,000 inserts/sec with a 1.2TB / 5 billion object DB.
There are a few ways to connect to MongoDB from .Net, but I don't have enough experience with that platform to know which is best:
Norm: http://wiki.github.com/atheken/NoRM/
MongoDB-CSharp: http://github.com/samus/mongodb-csharp
Simple-MongoDB: http://code.google.com/p/simple-mongodb/
Disclaimer: I work for 10gen on MongoDB so I am a bit biased.
I'm running a website for a jewellery wholesaler.
The prices on all their products are calculated using the current bullion metal fixes that are updated every night.
Currently, for the website, the calculations are worked out via a php include function, which works fine under current circumstances.
There are around 10,000 products, but the prices are calculated in real-time (ie when the web page is requested). The calculations are simple, but there are lots of them (Around 50+) and i'm worried that an increase in traffic may slow the current script down.
I'm redesigning the site and was wondering whether it would be beneficial to create a procedure in MySQL to do the calculations instead.
Is this likely to be faster that the current php script?
Anyone know any good reading reference on using procedures?
Here's a benchmark with stored procedure vs php.
http://mtocker.livejournal.com/45222.html
The stored procedure was slower by 10x.
You might also want to look at this:
http://www.tonymarston.net/php-mysql/stored-procedures-are-evil.html
If the reason you are thinking about this is due to performance and scalability, then I would recommend continuing the calculation in PHP.
The reason for this is that regardless whether there is a performance penalty in your PHP, when you are scaling your web application it is generally much easier to move to multiple web servers than multiple database servers. It is therefore preferable to do more calculation in PHP and less in MySQL.
Other than the performance aspect, I still generally prefer avoiding stored procedures in favour of having the logic in the application because
It can be less portable. Stored procedure add to the effort required to deploy a new instance of your application.
They are written in a different language than PHP, so a PHP developer may not find them easy to understand.
It can be difficult to have them kept in source control.
These problems can of course all be solved, without too much difficulty, but it all adds to the complexity overhead.
If it is absolutely necessary to update the prices on every page request and you're worried that the site will be getting a lot of traffic I wouldn't recommend stored procedures.
I'd recommend caching the information you use (and it is hard to elaborate without knowing how you're doing this) in memory (perhaps using memcached) and keep reading it from PHP.
I'll admit that I haven't done any benchmarking between stored procedures vs in-memory PHP performance but if the procedure doesn't directly affect your query, I recommend caching.
In short, keep them in php. Easier to maintain.
For the current site, there is unlikely to ever hit a performance problem where the difference between the speed of calc in php vs the speed of the calc in the database is ever noticeable. If you were then there is something fundamentally wrong with the code of the site. (This includes realtime currency conversions if being done).
Saying that, keeping the calc in PHP is usually preferred as it is easier to control and debug. It does require the web coders to know the database somewhat but that is normally not a problem. 90% of the code speed ups happen on 10% of the code and it would be easy enough for a dba to identify the queries causing db load if it ever happened.