I'm not a mongodb expert, so I'm a little unsure about server setup now.
I have a single instance running mongo3.0.2 with wiredtiger, accepting both read and write ops. It collects logs from client, so write load is decent. Once a day I want to process this logs and calculate some metrics using aggregation framework, data set to process is something like all logs from last month and all calculation takes about 5-6 hours.
I'm thinking about splitting write and read to avoid locks on my collections (server continues to write logs while i'm reading, newly written logs may match my queries, but i can skip them, because i don't need 100% accuracy).
In other words, i want to make a setup with a secondary for read, where replication is not performing continuously, but starts in a configured time or better is triggered before all read operations are started.
I'm making all my processing from node.js so one option i see here is to export data created in some period like [yesterday, today] and import it to read instance by myself and make calculations after import is done. I was looking on replica set and master/slave replication as possible setups but i didn't get how to config it to achieve the described scenario.
So maybe i wrong and miss something here? Are there any other options to achieve this?
Your idea of using a replica-set is flawed for several reasons.
First, a replica-set always replicates the whole mongod instance. You can't enable it for individual collections, and certainly not only for specific documents of a collection.
Second, deactivating replication and enabling it before you start your report generation is not a good idea either. When you enable replication, the new slave will not be immediately up-to-date. It will take a while until it has processed the changes since its last contact with the master. There is no way to tell how long this will take (you can check how far a secondary is behind the primary using rs.status() and comparing the secondaries optimeDate with its lastHeartbeat date).
But when you want to perform data-mining on a subset of your documents selected by timespan, there is another solution.
Transfer the documents you want to analyze to a new collection. You can do this with an aggregation pipeline consisting only of a $match which matches the documents from the last month followed by an $out. The out-operator specifies that the results of the aggregation are not sent to the application/shell, but instead written to a new collection (which is automatically emptied before this happens). You can then perform your reporting on the new collection without locking the actual one. It also has the advantage that you are now operating on a much smaller collection, so queries will be faster, especially those which can't use indexes. Also, your data won't change between your aggregations, so your reports won't have any inconsistencies between them due to data changing between them.
When you are certain that you will need a second server for report generation, you can still use replication and perform the aggregation on the secondary. However, I would really recommend you to build a proper replica-set (consisting of primary, secondary and an arbiter) and leave replication active at all times. Not only will that make sure that your data isn't outdated when you generate your reports, it also gives you the important benefit of automatic failover should your primary go down for some reason.
Related
In an NodeJS application I have to maintain a "who was online in the last N minutes" state. Since there is potentially thousands of online users - for performance reasons - I decided to not update my Postgresql user table for this task.
I choosed to use Redis to manage the online status. It's very easy and efficient.
But now I want to make complex queries to the user table, sorted by the online status.
I was thinking of creating a online table filled every minute from a Redis snapshot, but I'm not sure it's the best solution.
Following the table filling, will the next query referencing the online table take a big hit caused by the new indexes creation or loading?
Does anyone know a better solution?
I had to solve almost this exact same issue, but I took a different approach because I Didn't like the issues caused by trying to mix Redis and Postgres.
My solution was to collect the online data in a queue (Zero MQ in my case) but any queueing system should work, or a stream processing facility like Amazon Kinesis (The alternative I looked at.) I then inserted the data in batches into a second table (not the users table). I don't delete or update that table, only inserts and queries are allowed.
Doing things this way preserved the ability to do joins between the last online data and the users table without bogging down the database or creating many updates on the user tables. It has the side effect of giving us a lot of other useful data.
One thing to note that I have though about when thinking of other solutions to this problem is that your users table in transactional data(OLTP) while the latest online information is really analytics data (OLAP), so if you have a data warehouse, data lake, big data, or whatever term of the week you want to use for storing this type of data and querying against it that may be a better solution.
Background
We have recently started a "Big Data" project where we want to track what users are doing with our product - how often they are logging in, which features they are clicking on, etc - your basic user analytics stuff. We still don't know exactly what questions we will be asking, but most of it will be "how often did X occur over the last Y months?" type of thing, so we started storing the data sooner rather than later thinking we can always migrate, re-shape etc when we need to but if we don't store it it is gone forever.
We are now looking at what sorts of questions we can ask. In a typical RDBMS, this stage would consist of slicing and dicing the data in many different dimensions, exporting to Excel, producing graphs, looking for trends etc - it seems that for Cassandra, this is rather difficult to do.
Currently we are using Apache Spark, and submitting Spark SQL jobs to slice and dice the data. This actually works really well, and we are getting the data we need, but it is rather cumbersome as there doesn't seem to be any native API for Spark that we can connect to from our workstations, so we are stuck using the spark-submit script and a Spark app that wraps some SQL from the command line and outputs to a file which we then have to read.
The question
In a table (or Column Family) with ~30 columns running on 3 nodes with RF 2, how bad would it be to add an INDEX to every non-PK column, so that we could simply query it using CQL across any column? Would there be a horrendous impact on the performance of writes? Would there be a large increase in disk space usage?
The other option I have been investigating is using Triggers, so that for each row inserted, we populated another handful of tables (essentially, custom secondary index tables) - is this a more acceptable approach? Does anyone have any experience of the performance impact of Triggers?
Impact of adding more indexes:
This really depends on your data structure, distribution and how you access it; you were right before when you compared this process to RDMS. For Cassandra, it's best to define your queries first and then build the data model.
These guys have a nice write-up on the performance impacts of secondary indexes:
https://pantheon.io/blog/cassandra-scale-problem-secondary-indexes
The main impact (from the post) is that secondary indexes are local to each node, so to satisfy a query by indexed value, each node has to query its own records to build the final result set (as opposed to a primary key query where it is known exactly which node needs to be quired). So there's not just an impact on writes, but on read performance as well.
In terms of working out the performance on your data model, I'd recommend using the cassandra-stress tool; you can combine it with a data modeler tool that Datastax have built, to quickly generate profile yamls:
http://www.datastax.com/dev/blog/data-modeler
For example, I ran the basic stress profile without and then with secondary indexes on the default table, and the "with indexes" batch of writes took a little over 40% longer to complete. There was also an increase in GC operations / duration etc.
It may be too much turkey over the holidays, but I've been thinking about a potential problem that we could have with Couchbase.
Currently we paginate based on time, but I'm thinking a similar issue could occur with other values used for paging for example the atomic counter. I'll try to explain best I can, this would only occur in a load balanced environment.
For example say we have 4 servers load balanced and storing data to our Couchbase cluster. We sort our records based on timestamps currently. If any of the 4 servers writing the data starts to lag behind the others than our pagination would possibly be missing records when retrieving client side. A SQL DB auto-increment and timestamps for example can be created when the record is stored to the DB which will avoid similar issues. Using a NoSql DB like Couchbase you define the data you need to retrieve on before it is stored to the DB. So what I am getting at is if there is a delay in storing to the DB and you are retrieving in a pagination fashion while this delay has occurred, you run the real possibility of missing data. Since we are paging that data may never be viewed.
Interested in what other thoughts people have on this.
EDIT**
Response to Andrew:
Example a facebook or pintrest type app is storing data to a DB, they have many load balanced servers from the frontend writing to the db. If for some reason writing is delayed its a non issue with a SQL DB because a timestamp or auto increment happens when the data is actually stored to the DB. There will be no missing data when paging. asking for 1-7 will give you data that is only stored in the DB, 7-* will contain anything that is delayed because an auto-increment value has not been created for that record becuase it is not actually stored.
In Couchbase its different, you actually get your auto increment value (atomic counter) and then save it. So for example say a record is going to be stored as atomic counter number 4. For some reasons this is delayed in storing to the DB. Other servers are grabbing 5, 6, 7 and storing that data just fine. The client now asks for all data between 1 and 7, 4 is still not stored. Then the next paging request is 7 to *. 4 will never be viewed.
Is there a way around this? Can it be modelled differently in CB, or is this just a potential weakness in CB when needing to page results. As I mentioned are paging is timestamp sensitive.
Michael,
Couchbase is an eventually consistent database with respect to views. It is ACID with respect to documents. There are durability interfaces that let you manage this. This means that you can rest assured you won't lose data and that indexes will catch up eventually.
In my experience with Couchbase, you need to expect that the nodes will never be in-sync. There are many things the database is doing, such as compaction and replication. The most important thing you can do to enhance performance is to put your views on a separate spindle from the data. And you need to ensure that your main data spindles across your cluster can sustain between 3-4 times your ingestion bandwidth. Also, make sure your main document key hashes appropriately to distribute the load.
It sounds like you are discussing a situation where the data exists in your system for less time than it takes to be processed through the view system. If you are removing data that fast, you need either a bigger cluster or faster disk arrays. Of the two choices, I would expand the size of your cluster. I like to think of Couchbase as building a RAIS, Redundant Array of Independent Servers. By expanding the cluster, you reduce the coincidence of hotspots and gain disk bandwidth. My ideal node has two local drives, one each for data and views, and enough RAM for my working set.
Anon,
Andrew
Can CouchDB handle thousands of separate databases on the same machine?
Imagine you have a collection of BankTransactions. There are many thousands of records. (EDIT: not actually storing transactions--just think of a very large number of very small, frequently updating records. It's basically a join table from SQL-land.)
Each day you want a summary view of transactions that occurred only at your local bank branch. If all the records are in a single database, regenerating the view will process all of the transactions from all of the branches. This is a much bigger chunk of work, and unnecessary for the user who cares only about his particular subset of documents.
This makes it seem like each bank branch should be partitioned into its own database, in order for the views to be generated in smaller chunks, and independently of each other. But I've never heard of anyone doing this, and it seems like an anti-pattern (e.g. duplicating the same design document across thousands of different databases).
Is there a different way I should be modeling this problem? (Should the partitioning happen between separate machines, not separate databases on the same machine?) If not, can CouchDB handle the thousands of databases it will take to keep the partitions small?
(Thanks!)
[Warning, I'm assuming you're running this in some sort of production environment. Just go with the short answer if this is for a school or pet project.]
The short answer is "yes".
The longer answer is that there are some things you need to watch out for...
You're going to be playing whack-a-mole with a lot of system settings like max file descriptors.
You'll also be playing whack-a-mole with erlang vm settings.
CouchDB has a "max open databases" option. Increase this or you're going to have pending requests piling up.
It's going to be a PITA to aggregate multiple databases to generate reports. You can do it by polling each database's _changes feed, modifying the data, and then throwing it back into a central/aggregating database. The tooling to make this easier is just not there yet in CouchDB's API. Almost, but not quite.
However, the biggest problem that you're going to run into if you try to do this is that CouchDB does not horizontally scale [well] by itself. If you add more CouchDB servers they're all going to have duplicates of the data. Sure, your max open dbs count will scale linearly with each node added, but other things like view build time won't (ex., they'll all need to do their own view builds).
Whereas I've seen thousands of open databases on a BigCouch cluster. Anecdotally that's because of dynamo clustering: more nodes doing different things in parallel, versus walled off CouchDB servers replicating to one another.
Cheers.
I know this question is old, but wanted to note that now with more recent versions of CouchDB (3.0+), partitioned databases are supported, which addresses this situation.
So you can have a single database for transactions, and partition them by bank branch. You can then query all transactions as you would before, or query just for those from a specific branch, and only the shards where that branch's data is stored will be accessed.
Multiple databases are possible, but for most cases I think the aggregate database will actually give better performance to your branches. Keep in mind that you're only optimizing when a document is updated into the view; each document will only be parsed once per view.
For end-of-day polling in an aggregate database, the first branch will cause 100% of the new docs to be processed, and pay 100% of the delay. All other branches will pay 0%. So most branches benefit. For end-of-day polling in separate databases, all branches pay a portion of the penalty proportional to their volume, so most come out slightly behind.
For frequent view updates throughout the day, active branches prefer the aggregate and low-volume branches prefer separate. If one branch in 10 adds 99% of the documents, most of the update work will be done on other branch's polls, so 9 out of 10 prefer separate dbs.
If this latency matters, and assuming couch has some clock cycles going unused, you could write a 3-line loop/view/sleep shell script that updates some documents before any user is waiting.
I would add that having a large number of databases creates issues around compaction and replication. Not only do things like continuous replication need to be triggered on a per-database basis (meaning you will have to write custom logic to loop over all the databases), but they also spawn replication daemons per database. This can quickly become prohibitive.
I have an application that works as follows: Linux machines generate 28 different types of letter to customers. The letters must be sent in .docx (Microsoft Word format). A secretary maintains MS Word templates, which are automatically used as necessary. Changing from using MS Word is not an option.
To coordinate all this, document jobs are placed into a database table and a python program running on each of the windows machines polls the database frequently, locking out jobs and running them as necessary.
We use a central database table for the job information to coordinate different states ("new", "processing", "finished", "printed")... as well to give accurate status information.
Anyway, I don't like the clients polling the database frequently, seeing as they aren't working most of the time. Clients hpoll every 5 seconds.
To avoid polling, I kind of want a broadcast "there's some work to do" or "check your database for some work to do" message sent to all the client machines.
I think some kind of publish/subscribe message queue would be up to the job, but I don't want any massive extra complexity.
Is there a zero or near zero config/maintenance piece of software that would achieve this? What are the options?
X
Is there any objective evidence that any significant load is being put on the server? If it works, I'd make sure there's really a problem to solve here.
It must be nice to have everything running so smoothly that you're looking at things that might only possibly be improved!
Is there a zero or near zero config/maintenance piece of software that would achieve this? What are the options?
Possibly, but what you would save in configuration and implementation time would likely hurt performance more than your polling service ever could. SQL Server isn't made to do a push really (not easily anyway). There are things that you could use to push data out (replication service, log shipping - icky stuff), but they would be more complex and require more resources than your simple polling service. Some options would be:
some kind of trigger which runs your executable using command-line calls (sp_cmdshell)
using a COM object which SQL Server could open and run
using a SQL Agent job to run a VBScript (which would again be considered "polling")
These options are a bit ridiculous considering what you have already done is much simpler.
If you are worried about the polling service using too many cycles or something - you can always throttle it back - polling every minute, every 10 minutes, or even just once a day might be more appropriate - this would be a business decision, so go ask someone in the business how fast it needs to be.
Simple polling services are fairly common, because they are, well... simple. In addition they are also low overhead, remotely stable, and error-tolerant. The down side is that they can hammer the database into dust if not carefully controlled.
A message queue might work well, as they're usually setup to be able to block for a while without wasting resources. But with MySQL, I don't think that's an option.
If you just want to reduce load on the DB, you could create a table with a single row: the latest job ID. Then clients just need to compare that to their last ID to see if they need to run a full poll against the real table. This way the overhead should be greatly reduced, if it's an issue now.
Unlike Postgres and SQL Server (or object stores like CouchDb), MySQL does not emit database events. However there are some coding patterns you can use to simulate this.
If you have one or more tables that you wish to monitor, you can create triggers on these tables that add a row to a "changes" table that records a queue of events to process. Your triggers filter the subset of data changes that you care about and create records in your changes table for each event you wish to perform. Because this pattern queues and persists events it works well even when the workers that process these events have outages.
You might think that MyISAM is the best choice for the changes table since it's mostly performing writes (or even MEMORY if you don't need to persist the events between database server outages). However, keep in mind that both Memory and MEMORY and MyISAM have only full-table locks so your trigger on an InnoDB table might hit a bottle neck when performing an insert into a MEMORY and MyISAM table. You may also require InnoDB for the changes table if you're using a ON DELETE CASCADE with another InnoDB table (requires both tables to use the same engine).
You might also use SHOW TABLE STATUS to check the last update time of you changes table to check if there's something to perform. This feature wont work for InnoDB tables.
These articles describes in more depth some of alternative ways to implement queues in MySQL and even avoid polling!
How to notify event listeners in MySQL
How to implement a queue in SQL
5 subtle ways you're using MySQL as a queue, and why it'll bite you