Interested if anyone knows about what could be going on under the hood to cause the same query to fail at one Azure SQL performance level but work at a higher level?
The query does max out the server to 100% at the Standard level, but I would expect to get an out of memory related exception if this was the issue. But instead I get a
Cannot create a row of size 8075 which is greater than the allowable maximum row size of 8060
I am aware that the query needs to be optimized, but what I am interested in for the purposes of this question is what about bumping up to Premium 150DTU would suddenly make the same data not exceed the max row size?
I can make an educated guess as to what your problem is. When you change from one reservation size to another, the resources available to the optimizer changes. It believes it has more memory, specifically. Memory is a key component in costing queries in the query optimizer. The plan choice happening in the lower reservation size likely has a spool or sort that is trying to create an object in tempdb. The one in the higher reservation size is not. This it hitting a limitation in the storage engine since the intermediate table can not be materialized.
Without looking at the plan, it is not possible to say with certainty whether this is a requirement of the chosen query plan or merely an optimization. However, you can try using the NO_PERFORMANCE_SPOOL hint on the query to see if that makes it work on the smaller reservation size. (Given that it has less memory, I will guess that it is not the issue, however).
https://learn.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-query?view=sql-server-2017
(Now I am guessing with general advice since I don't know what kind of app you have but it based on the normal patterns I see regularly):
If your schema is really wide or poorly defined, please consider revising your table definition to reduce the size of the columns to the right minimum. For data warehousing applications, please consider using dimension tables + surrogate keys. If you are dumping text log files into SQL and then trying to distinct them, note that distinct is often going to imply a sort which could lead to this kind of issue if the row is too wide (as you are trying to use all columns in the key).
Best of luck on getting your app to work. SQL tends to work very well for you for relational apps where you think through the details of your schema and indexes a bit. In the future, please post a bit more detail about your query patterns, schema, and query plans so others can help you more precisely.
We are building cv scoring service, and we are using Postgres for making complex queries to find cv's that match vacancy best.
The problem is, that we use really complex set of heuristics to score cv to vacancy, and the average number of cvs to be scored per query is growing.
I want to put this kind of load outside of database, and looking for existing solutions for horizontal scaling such load.
Query should be executed in fraction of a second, there can be hundreds of concurrent queries. Each query scores on average 10k cvs. Each cv is like about 50 records in maybe 10 tables in its current relational form.
I want a clustered system to run each query in multiple parallel processes (on many servers) and return aggregated result. It should be fast and fault tolerant.
I was looking to Hadoop, but it looks like it is designed for batch processing, and not for realtime low latency load. There is Apache Storm, but it is designed for continous stream processing. So I am not shure :)
What kind of tool could will suit my needs?
Thank you!
Make sure you are not redoing work, if a cv has been scored tag it as scored and don't reprocess unless it's necessary.
Unless you are partitioning the data in postgres you might want to do that. Usually not all rows need to be accessed regularly.
Sounds like you want to primarily scale reads, in that case a postgres read-only cluster could be an option.
Take a look at Elasticsearch, it is designed to do weighted scoring, faceting, etc. It should also scale, haven't tried that myself though.
I would definitely start with 1 though, don't do work unless you have to.
I am currently addressing a situation where our web application receives at least a Million requests per 30 seconds. So these requests will lead to generating 3-5 Million row inserts between 5 tables. This is pretty heavy load to handle. Currently we are using multi threading to handle this situation (which is a bit faster but unable to get a better CPU throughput). However the load will definitely increase in future and we will have to account for that too. After 6 months from now we are looking at double the load size we are currently receiving and I am currently looking at a possible new solution that is scalable and should be easy enough to accommodate any further increase to this load.
Currently with multi threading we are making the whole debugging scenario quite complicated and sometimes we are having problem with tracing issues.
FYI we are already utilizing the SQL Builk Insert/Copy that is mentioned in this previous post
Sql server 2008 - performance tuning features for insert large amount of data
However I am looking for a more capable solution (which I think there should be one) that will address this situation.
Note: I am not looking for any code snippets or code examples. I am just looking for a big picture of a concept that I could possibly use and I am sure that I can take that further to an elegant solution :)
Also the solution should have a better utilization of the threads and processes. And I do not want my threads/processes to even wait to execute something because of some other resource.
Any suggestions will be deeply appreciated.
Update: Not every request will lead to an insert...however most of them will lead to some sql operation. The appliciation performs different types of transactions and these will lead to a lot of bulk sql operations. I am more concerned towards inserts and updates.
and these operations need not be real time there can be a bit lag...however processing them real time will be much helpful.
I think your problem looks more towards getting a better CPU throughput which will lead to a better performance. So I would probably look at something like an Asynchronous Processing where in a thread will never sit idle and you will probably have to maintain a queue in the form of a linked list or any other data structure that will suit your programming model.
The way this would work is your threads will try to perform a given job immediately and if there is anything that would stop them from doing it then they will push that job into the queue and these pushed items will be processed based on how it stores the items in the container/queue.
In your case since you are already using bulk sql operations you should be good to go with this strategy.
lemme know if this helps you.
Can you partition the database so that the inserts are spread around? How is this data used after insert? Is there a natural partion to the data by client or geography or some other factor?
Since you are using SQL server, I would suggest you get several of the books on high availability and high performance for SQL Server. The internals book muight help as well. Amazon has a bunch of these. This is a complex subject and requires too much depth for a simple answer on a bulletin board. But basically there are several keys to high performance design including hardware choices, partitioning, correct indexing, correct queries, etc. To do this effectively, you have to understand in depth what SQL Server does under the hood and how changes can make a big difference in performance.
Since you do not need to have your inserts/updates real time you might consider having two databases; one for reads and one for writes. Similar to having a OLTP db and an OLAP db:
Read Database:
Indexed as much as needed to maximize read performance.
Possibly denormalized if performance requires it.
Not always up to date.
Insert/Update database:
No indexes at all. This will help maximize insert/update performance
Try to normalize as much as possible.
Always up to date.
You would basically direct all insert/update actions to the Insert/Update db. You would then create a publication process that would move data over to the read database at certain time intervals. When I have seen this in the past the data is usually moved over on a nightly bases when few people will be using the site. There are a number of options for moving the data over, but I would start by looking at SSIS.
This will depend on your ability to do a few things:
have read data be up to one day out of date
complete your nightly Read db update process in a reasonable amount of time.
Im new to database design and need some guidance.
A lot of new data is inserted to my database throughout the day. (100k rows per day)
The data is never modified or deleted once it has been inserted.
How can I optimize this database for retrieval speed?
My ideas
Create two databases (and possible on different hard drives) and merge the two at night when traffic is low
Create some special indexes...
Your recommendation is highly appreciated.
UPDATE:
My database only has a single table.
100k/day is actually fairly low. 3M/month, 40M/year. You can store 10 years archive and not reach 1B rows.
The most important thing to choose in your design will be the clustered key(s). You need to make sure that they are narrow and can serve all the queries your application will normally use. Any query that will end up in table scan will completely trash your memory by fetching in the entire table. So, no surprises there, your driving factor in your design is the actual load you'll have: exactly what queries will you be running.
A common problem (more often neglected than not) with any high insert rate is that eventually every row inserted will have to be deleted. Not acknowledging this is a pipe dream. The proper strategy depends on many factors, but probably the best bet is on a sliding window partitioning scheme. See How to Implement an Automatic Sliding Window in a Partitioned Table. This cannot be some afterthought, the choice for how to remove data will permeate every aspect of your design and you better start making a strategy now.
The best tip I can give which all big sites use to speed up there website is:
CACHE CACHE CACHE
use redis/memcached to cache your data! Because memory is (blazingly)fast and disc I/O is expensive.
Queue writes
Also for extra performance you could queue up the writes in memory for a little while before flushing them to disc -> writting them to SQL database. Off course then you have the risk off losing data if you keep it in memory and your computer crashes or has power failure or something
Context missing
Also I don't think you gave us much context!
What I think is missing is:
architecture.
What kind of server are you having VPS/shared hosting.
What kind of Operating system does it have linux/windows/macosx
computer specifics like how much memory available, cpu etc.
a find your definition of data a bit vague. Could you not attach a diagram or something which explains your domain a little bit. For example something like
this using http://yuml.me/
Your requirements are way to general. For MS SQL server 100k (more or less "normal") records per days should not be a problem, if you have decent hardware. Obviously you want to write fast to the database, but you ask for optimization for retrieval performance. That does not match very well! ;-) Tuning a database is a special skill on its own. So you will never get the general answer you would like to have.
I have a database containing a single huge table. At the moment a query can take anything from 10 to 20 minutes and I need that to go down to 10 seconds. I have spent months trying different products like GridSQL. GridSQL works fine, but is using its own parser which does not have all the needed features. I have also optimized my database in various ways without getting the speedup I need.
I have a theory on how one could scale out queries, meaning that I utilize several nodes to run a single query in parallel. A precondition is that the data is partitioned (vertically), one partition placed on each node. The idea is to take an incoming SQL query and simply run it exactly like it is on all the nodes. When the results are returned to a coordinator node, the same query is run on the union of the resultsets. I realize that an aggregate function like average need to be rewritten into a count and sum to the nodes and that the coordinator divides the sum of the sums with the sum of the counts to get the average.
What kinds of problems could not easily be solved using this model. I believe one issue would be the count distinct function.
Edit: I am getting so many nice suggestions, but none have addressed the method.
It's a data volume problem, not necessarily an architecture problem.
Whether on 1 machine or 1000 machines, if you end up summarizing 1,000,000 rows, you're going to have problems.
Rather than normalizing you data, you need to de-normalize it.
You mention in a comment that your data base is "perfect for your purpose", when, obviously, it's not. It's too slow.
So, something has to give. Your perfect model isn't working, as you need to process too much data in too short of a time. Sounds like you need some higher level data sets than your raw data. Perhaps a data warehousing solution. Who knows, not enough information to really say.
But there are a lot of things you can do to satisfy a specific subset of queries with a good response time, while still allowing ad hoc queries that respond in "10-20 minutes".
Edit regarding comment:
I am not familiar with "GridSQL", or what it does.
If you send several, identical SQL queries to individual "shard" databases, each containing a subset, then the simple selection query will scale to the network (i.e. you will eventually become network bound to the controller), as this is a truly, parallel, stateless process.
The problem becomes, as you mentioned, the secondary processing, notably sorting and aggregates, as this can only be done on the final, "raw" result set.
That means that your controller ends up, inevitably, becoming your bottleneck and, in the end, regardless of how "scaled out" you are, you still have to contend with a data volume issue. If you send your query out to 1000 node and inevitably have to summarize or sort the 1000 row result set from each node, resulting in 1M rows, you still have a long result time and large data processing demand on a single machine.
I don't know what database you are using, and I don't know the specifics about individual databases, but you can see how if you actually partition your data across several disk spindles, and have a decent, modern, multi-core processor, the database implementation itself can handle much of this scaling in terms of parallel disk spindle requests for you. Which implementations actually DO do this, I can't say. I'm just suggesting that it's possible for them to (and some may well do this).
But, my general point, is if you are running, specifically, aggregates, then you are likely processing too much data if you're hitting the raw sources each time. If you analyze your queries, you may well be able to "pre-summarize" your data at various levels of granularity to help avoid the data saturation problem.
For example, if you are storing individual web hits, but are more interested in activity based on each hour of the day (rather than the subsecond data you may be logging), summarizing to the hour of the day alone can reduce your data demand dramatically.
So, scaling out can certainly help, but it may well not be the only solution to the problem, rather it would be a component. Data warehousing is designed to address these kinds of problems, but does not work well with "ad hoc" queries. Rather you need to have a reasonable idea of what kinds of queries you want to support and design it accordingly.
One huge table - can this be normalised at all?
If you are doing mostly select queries, have you considered either normalising to a data warehouse that you then query, or running analysis services and a cube to do your pre-processing for you?
From your question, what you are doing sounds like the sort of thing a cube is optimised for, and could be done without you having to write all the plumbing.
By trying custom solution (grid) you introduce a lot of complexity. Maybe, it's your only solution, but first did you try partitioning the table (native solution)?
I'd seriously be looking into an OLAP solution. The trick with the Cube is once built it can be queried in lots of ways that you may not have considered. And as #HLGEM mentioned, have you addressed indexing?
Even at in millions of rows, a good search should be logarithmic not linear. If you have even one query which results in a scan then your performance will be destroyed. We might need an example of your structure to see if we can help more?
I also agree fully with #Mason, have you profiled your query and investigated the query plan to see where your bottlenecks are. Adding nodes improving speed makes me think that your query might be CPU bound.
David,
Are you using all of the features of GridSQL? You can also use constraint exclusion partitioning, effectively breaking out your big table into several smaller tables. Depending on your WHERE clause, when the query is processed it may look at a lot less data and return results much faster.
Also, are you using multiple logical nodes per physical server? Configuring it that way can take advantage of otherwise idle cores.
If you monitor the servers during execution, is the bottleneck IO or CPU?
Also alluded to here is that you may want to roll up rows in your fact table into summary tables/cubes. I do not know enough about Tableau, will it automatically use the appropriate cube and drill down only when necessary? If so, it seems like you would get big gains doing something like this.
My guess (based on nothing but my gut) is that any gains you might see from parallelization will be eaten up by reaggregation and subsequent queries of the results. Further, I would think that writing might get more complicated with pk/fk/constraints. If this were my world, I would probably create many indexed views on top of my table (and other views) that optimized for the particular queries I need to execute (which I have worked with successfully on 10million+ row tables.)
If you run the incoming query, unpartitioned, on each node, why will any node finish before a single node running the same query would finish? Am I misunderstanding your execution plan?
I think this is, in part, going to depend on the nature of the queries you're executing and, in particular, how many rows contribute to the final result set. But surely you'll need to partition the query somehow among the nodes.
Your method to scale out queries works fine.
In fact, I've implemented such a method in:
http://code.google.com/p/shard-query
It uses a parser, but it supports most SQL constructs.
It doesn't yet support count(distinct expr) but this is doable and I plan to add support in the future.
I also have a tool called Flexviews (google for flexviews materialized views)
This tool lets you create materialized views (summary tables) which include various aggregate functions and joins.
Those tools combined together can yield massive scalability improvements for OLAP type queries.