I used Oracle for the half past year and learned some tricks of sql tuning,but now our DB is moving to greenplum and the project manager suggest us to change some of the codes that writted in Oracle sql for their efficiency or grammar.
I am curious that Are sql tuning ways same for different DB engine,like oracle,postgresql,mysql and so on？if yes or not,why?Any suggestion are welcomed!
some like:
in or exists
count(*) or count(column)
use index or not
use exact column instead of select *
For the most part the syntax that is used will remain the same, there may be small differences from one engine to another and you may run into different terms to achieve some of the more specific output or do more complex tasks. In order to achieve parity you will need to learn those new terms.
As far as tuning, this will vary from system to system. Specifically going from Oracle to Greenplum you are looking at moving from a database where efficiency in a query if often driven by dropping an index on the data. Where Greenplum is a parallel execution system where efficiency is gained by effectively distributing the data across multiple systems and querying them in parallel. In Greenplum indexing is an additional layer that usually does not add benefit, just additional overhead.
Even within a single system using changing the storage engine type can result in different ways to optimize a query. In practice queries are often moved to a new platform and work, but are far from optimal as they don't take advantage of optimizations of that platform. I would strongly suggest getting an understanding of the new platform and you should not go in assuming a query that is optimized for one platform is the optimal way to run it in another.
Getting specifics in why they differ requires someone to be an expert in bother to be able to compare both. I don't claim to know much of greenplum.
The basic principles which I would expect all developers to learn over time dont really change. But there are "quirks" of individual engines which make specific differences. From your question I would personally anticipate 1 and 4 to remain the same.
Indexing is something which does vary. For example the ability to use two indexes was not (is not?) Ubiquitous. I wouldn't like to guess which DBMS can / can't count columns from the second field in a composite index. And the way indexes are maintained is very different from one DBMS to the next.
From my own experience I've also seen differences caused by:
Different capabilities in the data access path. As an example, one optimisation is for a DBMS to create a bit map of rows (matching and not matching) the combine multiple bitmaps to select rows. A DBMS with this feature can use multiple indexes in a single query. One without it can't.
Availability of hints / lack of hints. Not all DBMS support them. I know they are very common in Oracle.
Different locking strategies. This is a big one and can really affect update and insert queries.
In some cases DBMS have very specific capabilities for certain types of data such as geographic data or searchable free text (natural language). In these cases the way of working with the data is entirely different from one DBMS to the next.
Related
I have a background that includes SQL Server and Informix database query optimisation (non big-data). I'm confident in how to maximise database performance on those systems. I've recently been working with BigQuery and big data (about 9+ months), and optimisation doesn't seem to work the same way. I've done some research and read some articles on optimisation, but I still need to better understand the basics of how to optimise on BigQuery.
In SQL Server/Informix, a lot of the time I would introduce a column index to speed up reads. BigQuery doesn't have indexes, so I've mainly been using clustering. When I've done benchmarking after introducing a cluster for a column that I thought should make a difference, I didn't see any significant change. I'm also not seeing a difference when I switch on query cacheing. This could be an unfortunate coincidence with the queries I've tried, or a mistaken perception, however with SQL Server/SQL Lite/Informix I'm used to seeing immediate significant improvement, consistently. Am I misunderstanding clustering (I know it's not exactly like an index, but I'm expecting it should work in a similar type of way), or could it just be that I've somehow been 'unlucky' with the optimisations.
And this is where the real point is. There's almost no such thing as being 'unlucky' with optimisation, but in a traditional RDBMS I would look at the execution plan and know exactly what I need to do to optimise, and find out exactly what's going on. With BigQuery, I can get the 'execution details', but it really isn't telling me much (at least that I can understand) about how to optimise, or how the query really breaks down.
Do I need a significantly different way of thinking about BigQuery? Or does it work in similar ways to an RDBMS, where I can consciously make the first JOINS eliminate as many records as possible, use 'where' clauses that focus on indexed columns, etc. etc.
I feel I haven't got the control to optimise like in a RDBMS, but I'm sure I'm missing a major point (or a few points!). What are the major strategies I should be looking at for BigQuery optimisation, and how can I understand exactly what's going on with queries? If anyone has any links to good documentation that would be fantastic - I'm yet to read something that makes me think "Aha, now I get it!".
It is absolutely a paradigm shift in how you think. You're right: you don't have hardly any control in execution. And you'll eventually come to appreciate that. You do have control over architecture, and that's where a lot of your wins will be. (As others mentioned in comments, the documentation is definitely helpful too.)
I've personally found that premature optimization is one of the biggest issues in BigQuery—often the things you do trying to make a query faster actually have a negative impact, because things like table scans are well optimized and there are internals that you can impact (like restructuring a query in a way that seems more optimal, but forces additional shuffles to disk for parallelization).
Some of the biggest areas our team HAS seem greatly improve performance are as follows:
Use semi-normalized (nested/repeated) schema when possible. By using nested STRUCT/ARRAY types in your schema, you ensure that the data is colocated with the parent record. You can basically think of these as tables within tables. The use of CROSS JOIN UNNEST() takes a little getting used to, but eliminating those joins makes a big difference (especially on large results).
Use partitioning/clustering on large datasets when possible. I know you mention this, just make sure that you're pruning what you can using _PARTITIONTIME when possible, and also using clutering keys that make sense for your data. Keep in mind that clustering basically sorts the storage order of the data, meaning that the optimizer knows it doesn't have to continue scanning if the criteria has been satisfied (so it doesn't help as much on low-cardinality values)
Use analytic window functions when possible. They're very well optimized, and you'll find that BigQuery's implementation is very mature. Often you can eliminate grouping this way, or filter our more of your data earlier in the process. Keep in mind that sometimes filtering data in derived tables or Common Table Expressions (CTEs/named WITH queries) earlier in the process can make a more deeply nested query perform better than trying to do everything in one flat layer.
Keep in mind that results for Views and Common Table Expressions (CTEs/named WITH queries) aren't materialized during execution. If you use the CTE multiple times, it will be executed multiple times. If you join the same View multiple times, it will be executed multiple times. This was hard for members of our team who came from the world of materialized views (although it looks like somethings in the works for that in BQ world since there's an unused materializedView property showing in the API).
Know how the query cache works. Unlike some platforms, the cache only stores the output of the outermost query, not its component parts. Because of this, only an identical query against unmodified tables/views will use the cache—and it will typically only persist for 24 hours. Note that if you use non-deterministic functions like NOW() and a host of other things, the results are non-cacheable. See details under the Limitations and Exceptions sections of the docs.
Materialize your own copies of expensive tables. We do this a lot, and use scheduled queries and scripts (API and CLI) to normalize and save a native table copy of our data. This allows very efficient processing and fast responses from our client dashboards as well as our own reporting queries. It's a pain, but it works well.
Hopefully that will give you some ideas, but also feel free to post queries on SO in the future that you're having a hard time optimizing. Folks around here are pretty helpful when you let them know what your data looks like and what you've already tried.
Good luck!
For testing purposes I require a large amount of queries.
Creating this manually is not an option, so I am searching a tool which will do this automatically.
Sadly, the only solution I found (sqlsmith), is limited to postgres and SQLite.
Are there any similar tools for SQL-Server?
"I do not know from what random place people will want to travel to a random other place, so instead, let's create roads for every possible combination of origin and destination".
That sounds kind of insane, doesn't it? The same applies to what you seem to be wanting to achieve. You basically are hoping to find a tool that generates random queries against your database so you can feed them to the tuning advisor, which will then suggest query optimization indexes for hypothetical queries.
If you want to performance tune your database, you should have a pretty good idea of the type of questions your users will be throwing at it, as well as the structure of your data. Typical questions that will help you get started would be things like:
What is the most common search my users would do against this table?
What criteria are they most likely to use?
Which columns are guaranteed or likely to contain unique data in every row?
Which columns will most likely have a low selectivity of data? (I.e. Male/Female)
are you looking for generate random data for multiple tables ? we generally use redgate data genearator tool for the same.
for SQL tuning purpose I would suggest
https://www.brentozar.com/blitzindex/
http://www.nguyenlamminhdieu.com/zone/213/news/vi-VN/zone/213/news/351-database-engine-tuning-advisor-in-sql-server.aspx
When people are comparing SQL and noSQL, and concluding the upsides and downsides of each one, what I never hear anyone talking about is the speed.
Isn't performing SQL queries generally faster than performing noSQL queries?
I mean, for me this would be a really obvious conclusion, because you should always be able to find something faster if you know the structure of your database than if you don't.
But people never seem to mention this, so I want to know if my conclusion is right or wrong.
People who tend to use noSQL use it specifically because it fits their use cases. Being divorced from normal RDBMS table relationships and constraints, as well as ACID-ity of data, it's very easy to make it run a lot faster.
Consider Twitter, which uses NoSQL because a user only does very limited things on site, or one exactly - tweet. And concurrency can be considered non-existent since (1) nobody else can modify your tweet and (2) you won't normally be simultaneously tweeting from multiple devices.
The definition of noSQL systems is a very broad one -- a database that doesn't use SQL / is not a RDBMS.
Therefore, the answer to your question is, in short: "it depends".
Some noSQL systems are basically just persistent key/value storages (like Project Voldemort). If your queries are of the type "look up the value for a given key", such a system will (or at least should be) faster that an RDBMS, because it only needs to have a much smaller feature set.
Another popular type of noSQL system is the document database (like CouchDB).
These databases have no predefined data structure.
Their speed advantage relies heavily on denormalization and creating a data layout that is tailored to the queries that you will run on it. For example, for a blog, you could save a blog post in a document together with its comments. This reduces the need for joins and lookups, making your queries faster, but it also could reduce your flexibility regarding queries.
As Einstein would say, speed is relative.
If you need to store a master/detail simple application (like a shopping cart), you would need to do several Insert statements in your SQL application, also you will get a Data set of information when you do a query to get the purchase, if you're using NoSQL, and you're using it well, then you would have all the data for a single order in one simple "record" (document if you use the terms of NoSQL databases like djondb).
So, I really think that the performance of an application can be measured by the number of things it need to do to achieve a single requirement, if you need to do several Inserts to store an order and you only need one simple Insert in a database like djondb then the performance will be 10x faster in the NoSQL world, just because you're using 10 times less calls to the database layer, that's it.
To illustrate my point let me link an example I wrote sometime ago about the differences between NoSQL and SQL data models approach: https://web.archive.org/web/20160510045647/http://djondb.com/blog/nosql-masterdetail-sample/, I know it's a self reference, but basically I wrote it to address this question which I found it's the most challenging question a RDBMS guy could have and it's always a good way to explain why NoSQL is so different from SQL world, and why it will achieve better performance anytime, not because we use "nasa" technology, it's because NoSQL will let the developer do less... and get more, and less code = greater performance.
The answer is: it depends. Generally speaking, the objective of NoSQL DATABASES (no "queries") is scalability. RDBMS usually have some hard limits at some point (I'm talking about millons and millons of rows) where you could not scale any more by traditional means (Replication, clustering, partitioning), and you need something more because your needs keep growing. Or even if you manage to scale, the overall setup is quite complicated. Or you can scale reads, but not writes.
And the queries depends on the particular implementation of your server, the type of query you are doing, the columns in the table, etc... remember that queries are just one part of the RDBMS.
query time of relational database like SQL for 1000 person data is 2000 ms and graph database like neo4j
is 2ms .if you crate more node 1000000 speed stable 2 ms
I figure there has to be a specific design reason why you can't write a query like the following one:
select
(select column_name
from information_schema
where column_name not like '%rate%'
and table_name = 'Fixed_Income')
from Fixed_Income
and instead have to resort to dynamic SQL.
Anyone knows what that reason is? I tried Googling it, but all the hits were cries for help in solving the problem -- meaning it's a pretty widespread need and not well understood.
The reason is that the query optimizer needs to know the exact schema objects you are referring to at compile time. It needs them to optimize the query. You wouldn't believe how slow the RDBMS would be without having this information available to the query optimizer.
It's a little like the performance difference of static vs. dynamic typing in practice: There is usually a non-trivial difference (I'm thinking just about mainstream languages here). The compiler can exploit the static information to generate great code.
Even if this feature was present, it would be implemented by first computing the table and column names and then doing a standard "static" query planning.
You ask a very interesting question.
The "relational" in "relational algebra" refers to name-value pairs, not to relationships between tables. In relational algebra, there is no requirement that all records in a set (table) have the same columns.
My best guess is that the limitation is related to the idea of entity-relationship diagrams comes into play. A database is designed around tables, and these tables have relationships to each other. The choice of a relational database for data storage and access was specifically when the data could be stored this way. Knowing the entities and their attributes suggests a static form of the data and hence static references in queries.
In addition, SQL as a language is a declarative language rather than a procedural language. This suggests -- but does not impose -- a compilation step separate from the running of the query. In general, the SQL engine does the following (at a very high level):
Compiles the query, generally into some sort of data flow process.
Optimizes the data flow process. (Typically part of the compilation process.)
Runs the query.
The first two result in what is called "the query plan". You really cannot do optimization, though, unless you know about the objects you are operating on. So, dynamically choosing tables and columns means that optimization would be part of running the query rather than compiling it.
Finally, some databases like SQL Server support dynamic SQL. This allows you to build strings that get compiled and run at the same time. This is very useful for complex decision support queries. It is not recommended when you need fast transaction throughput, because the overhead for compilation is too high relative to the query.
I've been reading a little about temporary tables in MySQL but I'm an admitted newbie when it comes to databases in general and MySQL in particular. I've looked at some examples and the MySQL documentation on how to create a temporary table, but I'm trying to determine just how temporary tables might benefit my applications and I guess secondly what sorts of issues I can run into. Granted, each situation is different, but I guess what I'm looking for is some general advice on the topic.
I did a little googling but didn't find exactly what I was looking for on the topic. If you have any experience with this, I'd love to hear about it.
Thanks,
Matt
Temporary tables are often valuable when you have a fairly complicated SELECT you want to perform and then perform a bunch of queries on that...
You can do something like:
CREATE TEMPORARY TABLE myTopCustomers
SELECT customers.*,count(*) num from customers join purchases using(customerID)
join items using(itemID) GROUP BY customers.ID HAVING num > 10;
And then do a bunch of queries against myTopCustomers without having to do the joins to purchases and items on each query. Then when your application no longer needs the database handle, no cleanup needs to be done.
Almost always you'll see temporary tables used for derived tables that were expensive to create.
First a disclaimer - my job is reporting so I wind up with far more complex queries than any normal developer would. If you're writing a simple CRUD (Create Read Update Delete) application (this would be most web applications) then you really don't want to write complex queries, and you are probably doing something wrong if you need to create temporary tables.
That said, I use temporary tables in Postgres for a number of purposes, and most will translate to MySQL. I use them to break up complex queries into a series of individually understandable pieces. I use them for consistency - by generating a complex report through a series of queries, and I can then offload some of those queries into modules I use in multiple places, I can make sure that different reports are consistent with each other. (And make sure that if I need to fix something, I only need to fix it once.) And, rarely, I deliberately use them to force a specific query plan. (Don't try this unless you really understand what you are doing!)
So I think temp tables are great. But that said, it is very important for you to understand that databases generally come in two flavors. The first is optimized for pumping out lots of small transactions, and the other is optimized for pumping out a smaller number of complex reports. The two types need to be tuned differently, and a complex report run on a transactional database runs the risk of blocking transactions (and therefore making web pages not return quickly). Therefore you generally don't want to avoid using one database for both purposes.
My guess is that you're writing a web application that needs a transactional database. In that case, you shouldn't use temp tables. And if you do need complex reports generated from your transactional data, a recommended best practice is to take regular (eg daily) backups, restore them on another machine, then run reports against that machine.
The best place to use temporary tables is when you need to pull a bunch of data from multiple tables, do some work on that data, and then combine everything to one result set.
In MS SQL, Temporary tables should also be used in place of cursors whenever possible because of the speed and resource impact associated with cursors.
If you are new to databases, there are some good books by Joe Kelko that review best practices for ANSI SQL. SQL For Smarties will describe in great detail the use of temp table, impact of indexes, where clauses, etc. It's a great reference book with in depth detail.
I've used them in the past when I needed to create evaluated data. That was before the time of views and sub selects in MySQL though and I generally use those now where I would have needed a temporary table. The only time I might use them is if the evaluated data took a long time to create.
I haven't done them in MySQL, but I've done them on other databases (Oracle, SQL Server, etc).
Among other tasks, temporary tables provide a way for you to create a queryable (and returnable, say from a sproc) dataset that's purpose-built. Let's say you have several tables of figures -- you can use a temporary table to roll those figures up to nice, clean totals (or other math), then join that temp table to others in your schema for final output. (An example of this, in one of my projects, is calculating how many scheduled calls a given sales-related employee must make per week, bi-weekly, monthly, etc.)
I also often use them as a means of "tilting" the data -- turning columns to rows, etc. They're good for advanced data processing -- but only use them when you need to. (My golden rule, as always, applies: If you don't know why you're using x, and you don't know how x works, then you probably shouldn't use it.)
Generally, I wind up using them most in sprocs, where complex data processing is needed. I'd love to give a concrete example, but mine would be in T-SQL (as opposed to MySQL's more standard SQL), and also they're all client/production code which I can't share. I'm sure someone else here on SO will pick up and provide some genuine sample code; this was just to help you get the gist of what problem domain temp tables address.