Optimizing Sql Transactions (large single transaction vs many small ones) - mysql

I'm working on a webserver. I can have an endpoint that compiles data in multiple transactions, or all in a single transaction. Which would be faster?/
Better?

The answer: It depends on the amount of data you would expect your database to return.
A: A lot of data being returned (Thousands, millions):
Suppose you are doing the next Facebook. If you are about to fetch a really enormous amount of data (2 millions of email addresses) it would probably be better to use some kind of "pagination" and fire a query every few seconds or minutes. You wouldn't want a query which waits for 10 minutes in order to get your results and keep the entire server busy.
B: Small or moderate amount of data being returned
Or, if you are about to fetch some moderate amount of data (300 cities, 523 employees and 43 phones) then you wouldn't want wasting transaction times by executing a separate SQL query for cities, employees and phones and try to use as few separate queries, as possible. This means probably using a lot of JOINs.

Related

Running a cron to update 1 million records in every hour fails

We have an E-commerce system with more than 1 million users with a total or 4 to 5 million records in order table. We use codeigniter framework as back end and Mysql as database.
Due to this excessive number of users and purchases, we use cron jobs to update the order details and referral bonus points in every hour to make the things work.
Now we have a situation that these data updates exceeds one hour and the next batch of updates reach before finishing the previous one, there by leading into a deadlock and failure of the system.
I'd like to know about the different possible architectural and database scaling options and suggestions to get rid of this situation. We are using only the monolithic architecture to run this application.
Don't use cron. Have a single process that starts over when it finishes. If one pass lasts more than an hour, the next one will start late. (Checking PROCESSLIST is clumsy and error-prone. OTOH, this continually-running approach needs a "keep-alive" cronjob.)
Don't UPDATE millions of rows. Instead, find a way to put the desired info in a separate table that the user joins to. Presumably, that extra table would only 1 row (if everyone is controlled by the same game) or a small number of rows (if there are only a small number of patterns to handle).
Do have the slowlog turned on, with a small value for long_query_time (possibly "1.0", maybe lower). Use pt-query-digest to summarize it to find the "worst" queries. Then we can help you make them take less time, thereby helping to calm your busy system and improve the 'user experience'.
Do use batched INSERT. (A one INSERT with 100 rows runs about 10 times as fast as 100 single-row INSERTs.) Batching UPDATEs is tricky, but can be done with IODKU.
Do use batches of 100-1000 rows. (This is somewhat optimal considering the various things that can happen.)
Do use transactions judiciously. Do check for errors (including deadlocks) at every step.
Do tell us what you are doing in the hourly update. We might be able to provide more targeted advice than that 15-year-old book.
Do realize that you have scaled beyond the capabilities of the typical 3rd-party package. That is, you will have to learn the details of SQL.
I have some ideas here for you - mixed up with some questions.
Assuming you are limited in what you can do (i.e. you can't re-architect you way out of this) and that the database can't be tuned further:
Make the list of records to be processed as small as possible
i.e. Does the job have to run over all records? These 4-5 million records - are they all active orders, or that's how many you have in total for all time? Obviously just process the bare minimum.
Split and parallel process
You mentioned "batches" but never explained what that meant - can you elaborate?
Can you get multiple instances of the cron job to run at once, each covering a different segment of the records?
Multi-Record Operations
The easy (lazy) way to program updates is to do it in a loop that iterates through each record and processes it individually, but relational databases can do updates over multiple records at once. I'm pretty sure there's a proper term for that but I can't recall it. Are you processing each row individually or doing multi-record updates?
How does the cron job query the database? Have you hand-crafted the most efficient queries possible, or are you using some ORM / framework to do stuff for you?

Database design for heavy timed data logging - Car Tracking System

I am a making a car tracking system and i want to store data that each car sends after every 5 seconds in a MySql database. Assuming that i have 1000 cars transmitting data to my system after 5 seconds, and the data is stored in one table. At some point i would want to query this table to generate reports for specific vehicle. I am confused between logging all the vehicles data in one table or creating a table for each vehicle (1000 tables). Which is more efficient?
OK 86400 seconds per day / 5 = 17280 records per car and day.
Will result in 17,280,000 records per day. This is not an issue for MYSQL in general.
And a good designed table will be easy to query.
If you go for one table for each car - what is, when there will be 2000 cars in future.
But the question is also: how long do you like to store the data?
It is easy to calculate when your database is 200 GB, 800GB, 2TB,....
One table, not one table per car. A database with 1000 tables will be a dumpster fire when you try to back it up or maintain it.
Keep the rows of that table as short as you possibly can; it will have many records.
Index that table both on timestamp and on (car_id, timestamp) . The second index will allow you to report on individual cars efficiently.
Read https://use-the-index-luke.com/
This is the "tip of the iceberg". There are about 5 threads here and on dba.stackexchange relating to tracking cars/trucks. Here are some further tips.
Keep datatypes as small as possible. Your table(s) will become huge -- threatening to overflow the disk, and slowing down queries due to "bulky rows mean that fewer rows can be cached in RAM".
Do you keep the "same" info for a car that is sitting idle overnight? Think of how much disk space this is taking.
If you are using HDD disks, plain on 100 INSERTs/second before you need to do some redesign of the ingestion process. (1000/sec for SSDs.) There are techniques that can give you 10x, maybe 100x, but you must apply them.
Will you be having several servers collecting the data, then doing simple inserts into the database? My point is that that may be your first bottleneck.
PRIMARY KEY(car_id, ...) so that accessing data for one car is efficient.
Today, you say the data will be kept forever. But have you computed how big your disk will need to be?
One way to shrink the data drastically is to consolidate "old" data into, say, 1-minute intervals after, say, one month. Start thinking about what you want to keep. For example: min/max/avg speed, not just instantaneous speed. Have an extra record when any significant change occurs (engine on; engine off; airbag deployed; etc)
(I probably have more tips.)

Best and most efficient way for ELO-score calculation for users in database

I'm having a hard time wrapping my head around the issue of an ELO-score-like calculation for a large amount of users on our platform.
For example. For every user in a large set of users, a complex formule, based on variable amounts of "things done", will result in a score for each user for a match-making-like principle.
For our situation, it's based on the amount of posts posted, connections accepted, messages sent, amount of sessions in a time period of one month, .. other things done etc.
I had two ideas to go about doing this:
Real-time: On every post, message, .. run the formula for that user
Once a week: Run the script to calculate everything for all users.
The concerns about these two I have:
Real-time: Would be an overkill of queries and calculations for each action a user performs. If let's say, 500 users are active, all of them are performing actions, the database would be having a hard time I think. There would them also run a script to re-calculate the score for inactive users (to lower their score)
Once a week: If we have for example 5.000 users (for our first phase), than that would result into running the calculation formula 5.000 times and could take a long time and will increase in time when more users join.
The calculation-queries for a single variable in a the entire formula of about 12 variables are mostly a simple 'COUNT FROM table', but a few are like counting "all connections of my connections" which takes a few joins.
I started with "logging" every action into a table for this purpose, just the counter values and increase/decrease them with every action and running the formula with these values (a record per week). This works but can't be applied for every variable (like the connections of connections).
Note: Our server-side is based on PHP with MySQL.
We're also running Redis, but I'm not sure if this could improve those bits and pieces.
We have the option to export/push data to other servers/databases if needed.
My main example is the app 'Tinder' which uses a sort-like algorithm for match making (maybe with less complex data variables because they're not using groups and communities that you can join)
I'm wondering if they run that real-time on every swipe, every setting change, .. or if they have like a script that runs continiously for a small batch of users each time.
Where it all comes down to. What would be the most efficient/non-database-table-locking way to do this, with keeping the idea in mind that there will be a moment that we're having 50.000 users for example?
The way I would handle this:
Implement the realtime algorithm.
Measure. Is it actually slow? Try optimizing
Still slow? Move the algorithm to a separate asynchronous process. Have the process run whenever there's an update. Really this is the same thing as 1, but it doesn't slow down PHP requests and if it gets busy, it can take more time to catch up.
Still slow? Now you might be able to optimize by batching several changes.
If you have 5000 users right now, make sure it runs well with 5000 users. You're not going to grow to 50.000 overnight, so adjust and invest in this as your problem changes. You might be surprised where your performance problems are.
Measuring is key though. If you really want to support 50K users right now, simulate and measure.
I suspect you should use the database as the "source of truth" aka "persistent storage".
Then fetch whatever is needed from the dataset when you update the ratings. Even lots of games by 5000 players should not take more than a few seconds to fetch and compute on.
Bottom line: Implement "realtime"; come back with table schema and SELECTs if you find that the table fetching is a significant fraction of the total time. Do the "math" in a programming language, not SQL.

Doing SUM() and GROUP BY over millions of rows on mysql

I have this query which only runs once per request.
SELECT SUM(numberColumn) AS total, groupColumn
FROM myTable
WHERE dateColumn < ? AND categoryColumn = ?
GROUP BY groupColumn
HAVING total > 0
myTable has less than a dozen columns and can grow up to 5 millions of rows, but more likely about 2 millions in production. All columns used in the query are numbers, except for dateColumn, and there are indexes on dateColumn and categoryColumn.
Would it be reasonble to expect this query to run in under 5 seconds with 5 million rows on most modern servers if the database is properly optimized?
The reason I'm asking is that we don't have 5 millions of data and we won't even hit 2 millions within the next few years, if the query doesn't run in under 5 seconds then, it's hard to know where the problem lies. Would it be because the query is not suitable for a large table, or the database isn't optimized, or the server isn't powerful enough? Basically, I'd like to know whether using SUM() and GROUP BY over a large table is reasonable.
Thanks.
As people in comments under your question suggested, the easiest way to verify is to generate random data and test query execution time. Please note that using clustered index on dateColumn can significantly change execution times due to the fact, that with "<" condition only subset of continuous disk data is retrieved in order to calculate sums.
If you are at the beginning of the process of development, I'd suggest concentrating not on the structure of table and indexes that collects data - but rather what do you expect to need to retrieve from the table in the future. I can share my own experience with presenting website administrator with web usage statistics. I had several webpages being requested from server, each of them falling into one on more "categories". My first approach was to collect each request in log table with some indexes, but the table grew much larger than I had at first estimated. :-) Due to the fact that statistics where analyzed in constant groups (weekly, monthly, and yearly) I decided to create addidtional table that was aggregating requests in predefined week/month/year grops. Each request incremented relevant columns - columns were refering to my "categories" . This broke some normalization rules, but allowed me to calculate statistics in a blink of an eye.
An important question is the dateColumn < ? condition. I am guessing it is filtering records that are out of date. It doesn't really matter how many records there are in the table. What matters is how much records this condition cuts down.
Having aggressive filtering by date combined with partitioning the table by date can give you amazing performance on ridiculously large tables.
As a side note, if you are not expecting to hit this much data in many years to come, don't bother solving it. Your business requirements may change a dozen times by then, together with the architecture, db layout, design and implementation details. planning ahead is great but sometimes you want to give a good enough solution as soon as possible and handle the future painful issues in the next release..

Right design for MySQL database

I want to build a MySQL database for storing the ranking of a game every 1h.
Since this database will become quite large in a short time, I figured it's important to have a proper design. Therefor some advice would be gratefully appreciated.
In order to keep it as small as possible, I decided to log only the first 1500 positions of the ranking. Every ranking of a player holds the following values:
ranking position, playername, location, coordinates, alliance, race, level1, level2, points1, points2, points3, points4, points5, points6, date/time
My approach was to simply grab all values of each top 1500 player every hour by a php script and insert them into the MySQL as one row. So every day the MySQL will grow 36,000 rows. I will have a second script that deletes every row that is older than 28 days, otherwise the database would get insanely huge. Both scripts will run as a cronjob.
The following queries will be performed on this data:
The most important one is simply the query for a certain name. It should return all stats for the player for every hour as an array.
The second is a query in which all players have to be returned that didn't gain points1 during a certain time period from the latest entry. This should return a list of players that didn't gain points (for the last 24h for example).
The third is a query in which all players should be listed that lost a certain amount or more points2 in a certain time period from the latest entry.
The queries shouldn't take a lifetime, so I thought I should probably index playernames, points1 and points2.
Is my approach to this acceptable or will I run into a performance/handling disaster? Is there maybe a better way of doing this?
Here is where you risk a performance problem:
Your indexes will speed up your reads, but will considerably slow down your writes. Especially since your DB will have over 1 million rows in that one table at any given time. Since your writes are happening via cron, you should be okay as long as you insert your 1500 rows in batches rather than one round trip to the DB for every row. I'd also look into query compiling so that you save that overhead as well.
Ranhiru Cooray is correct, you should only store data like the player name once in the DB. Create a players table and use the primary key to reference the player in your ranking table. The same will go for location, alliance and race. I'm guessing that those are more or less enumerated values that you can store in another table to normalize your design and be returned in your results with appropriates JOINs. Normalizing your data will reduce the amount of redundant information in your database which will decrease it's size and increase it's performance.
Your design may also be flawed in your ranking position. Can that not be calculated by the DB when you select your rows? If not, can it be done by PHP? It's the same as with invoice tables, you never store the invoice total because it is redundant. The items/pricing/etc can be used to calculate the order totals.
With all the adding/deleting, I'd be sure to run OPTIMIZE frequently and keep good backups. MySQL tables---if using MyISAM---can become corrupted easily in high writing/deleting scenarios. InnoDB tends to fair a little better in those situations.
Those are some things to think about. Hope it helps.