Introduction
For greater than a decade now, the Hive desk format has been a ubiquitous presence within the massive knowledge ecosystem, managing petabytes of information with outstanding effectivity and scale. However as the information volumes, knowledge selection, and knowledge utilization grows, customers face many challenges when utilizing Hive tables due to its antiquated directory-based desk format. Among the widespread points embrace constrained schema evolution, static partitioning of information, and lengthy planning time due to S3 listing listings.
Apache Iceberg is a contemporary desk format that not solely addresses these issues but additionally provides extra options like time journey, partition evolution, desk versioning, schema evolution, robust consistency ensures, object retailer file structure (the power to distribute recordsdata current in a single logical partition throughout many prefixes to keep away from object retailer throttling), hidden partitioning (customers don’t need to be intimately conscious of partitioning), and extra. Subsequently, Apache Iceberg desk format is poised to switch the standard Hive desk format within the coming years.
Nonetheless, as there are already 25 million terabytes of information saved within the Hive desk format, migrating current tables within the Hive desk format into the Iceberg desk format is critical for efficiency and price. Relying on the dimensions and utilization patterns of the information, a number of completely different methods may very well be pursued to attain a profitable migration. On this weblog, I’ll describe a couple of methods one might undertake for numerous use instances. Whereas these directions are carried out for Cloudera Knowledge Platform (CDP), Cloudera Knowledge Engineering, and Cloudera Knowledge Warehouse, one can extrapolate them simply to different providers and different use instances as nicely.
There are few situations that one may encounter. A number of of those use instances may suit your workload and also you may be capable of combine and match the potential options offered to fit your wants. They’re meant to be a common information. In all of the use instances we try emigrate a desk named “occasions.”
Method 1
You have got the power to cease your shoppers from writing to the respective Hive desk throughout the period of your migration. That is perfect as a result of this may imply that you simply don’t have to switch any of your consumer code. Typically that is the one alternative accessible you probably have a whole bunch of shoppers that may probably write to a desk. It may very well be a lot simpler to easily cease all these jobs slightly than permitting them to proceed throughout the migration course of.
In-place desk migration
Resolution 1A: utilizing Spark’s migrate process
Iceberg’s Spark extensions present an in-built process referred to as “migrate” emigrate an current desk from Hive desk format to Iceberg desk format. In addition they present a “snapshot” process that creates an Iceberg desk with a distinct identify with the identical underlying knowledge. You can first create a snapshot desk, run sanity checks on the snapshot desk, and make sure that all the pieces is so as.
As soon as you’re happy you’ll be able to drop the snapshot desk and proceed with the migration utilizing the migrate process. Needless to say the migrate process creates a backup desk named “events__BACKUP__.” As of this writing, the “__BACKUP__” suffix is hardcoded. There may be an effort underway to let the person move a customized backup suffix sooner or later.
Needless to say each the migrate and snapshot procedures don’t modify the underlying knowledge: they carry out in-place migration. They merely learn the underlying knowledge (not even full learn, they simply learn the parquet headers) and create corresponding Iceberg metadata recordsdata. For the reason that underlying knowledge recordsdata will not be modified, it’s possible you’ll not be capable of take full benefit of the advantages supplied by Iceberg straight away. You can optimize your desk now or at a later stage utilizing the “rewrite_data_files” process. This will likely be mentioned in a later weblog. Now let’s talk about the professionals and cons of this method.
PROS:
- Can do migration in phases: first do the migration after which perform the optimization later utilizing rewrite_data_files process (weblog to comply with).
- Comparatively quick because the underlying knowledge recordsdata are saved in place. You don’t have to fret about creating a short lived desk and swapping it later. The process will try this for you atomically as soon as the migration is completed.
- Since a Hive backup is on the market one can revert the change completely by dropping the newly created Iceberg desk and by renaming the Hive backup desk (__backup__) desk to its authentic identify.
CONS:
- If the underlying knowledge just isn’t optimized, or has a number of small recordsdata, these disadvantages may very well be carried ahead to the Iceberg desk as nicely. Question engines (Impala, Hive, Spark) may mitigate a few of these issues through the use of Iceberg’s metadata recordsdata. The underlying knowledge file areas is not going to change. So if the prefixes of the file path are widespread throughout a number of recordsdata we could proceed to endure from S3 throttling (see Object Retailer File Layout to see tips on how to configure it correctly.) In CDP we solely assist migrating exterior tables. Hive managed tables can’t be migrated. Additionally, the underlying file format for the desk needs to be certainly one of avro, orc, or parquet.
Notice: There may be additionally a SparkAction within the JAVA API.
Resolution 1B: utilizing Hive’s “ALTER TABLE” command
Cloudera applied a straightforward technique to do the migration in Hive. All you need to do is to change the desk properties to set the storage handler to “HiveIcebergStorageHandler.”
The professionals and cons of this method are basically the identical as Resolution 1B. The migration is completed in place and the underlying knowledge recordsdata will not be modified. Hive creates Iceberg’s metadata recordsdata for a similar actual desk.
Shadow desk migration
Resolution 1C: utilizing the CTAS assertion
This answer is most generic and it might probably be used with any processing engine (Spark/Hive/Impala) that helps SQL-like syntax.
You may run fundamental sanity checks on the information to see if the newly created desk is sound.
As soon as you’re happy along with your sanity checking you would rename your “occasions” desk to a “backup_events” desk after which rename your “iceberg_events” to “occasions.” Needless to say in some instances the rename operation may set off a listing rename of the underlying knowledge listing. If that’s the case and your underlying knowledge retailer is an object retailer like S3, that may set off a full copy of your knowledge and may very well be very costly. If whereas creating the Iceberg desk the placement clause is specified, then the renaming operation of the Iceberg desk is not going to trigger the underlying knowledge recordsdata to maneuver. The identify will change solely within the Hive metastore. The identical applies for Hive tables as nicely. In case your authentic Hive desk was not created with the placement clause specified, then the rename to backup will set off a listing rename. In that case, In case your filesystem is object retailer based mostly, then it may be greatest to drop it altogether. Given the nuances round desk rename it’s crucial to check with dummy tables in your system and test that you’re seeing your required habits earlier than you carry out these operations on crucial tables.
You may drop your “backup_events” if you want.
Your shoppers can now resume their learn/write operations on the “occasions” they usually don’t even have to know that the underlying desk format has modified. Now let’s talk about the professionals and cons of this method.
PROS:
- The newly created knowledge is nicely optimized for Iceberg and the information will likely be distributed nicely.
- Any current small recordsdata will likely be coalesced robotically.
- Frequent process throughout all of the engines.
- The newly created knowledge recordsdata might benefit from Iceberg’s Object Retailer File Structure, in order that the file paths have completely different prefixes, thus decreasing object retailer throttling. Please see the linked documentation to see tips on how to benefit from this function.
- This method just isn’t essentially restricted to migrating a Hive desk. One might use the identical method emigrate tables accessible in any processing engine like Delta, Hudi, and many others.
- You may change the information format say from “orc” to “parquet.’’
CONS
- It will set off a full learn and write of the information and it may be an costly operation.
- Your complete knowledge set will likely be duplicated. You could have adequate cupboard space accessible. This shouldn’t be an issue in a public cloud backed by an object retailer.
Method 2
You don’t have the posh of lengthy downtime to do your migration. You wish to let your shoppers or jobs proceed writing the information to the desk. This requires some planning and testing, however is feasible with some caveats. Right here is a method you are able to do it with Spark. You may probably extrapolate the concepts offered to different engines.
- Create an Iceberg desk with the specified properties. Needless to say you need to hold the partitioning scheme the identical for this to work accurately.
- Modify your shoppers or jobs to write down to each tables in order that they write to the “iceberg_events” desk and “occasions” desk. However for now, they solely learn from the “occasions” desk. Seize the timestamp from which your shoppers began writing to each of the tables.
- You programmatically record all of the recordsdata within the Hive desk that have been inserted earlier than the timestamp you captured in step 2.
- Add all of the recordsdata captured in step 3 to the Iceberg desk utilizing the “add_files” process. The “add_files” process will merely add the file to your Iceberg desk. You additionally may be capable of benefit from your desk’s partitioning scheme to skip step 3 completely and add recordsdata to your newly created Iceberg desk utilizing the “add_files” process.
- For those who don’t have entry to Spark you may merely learn every of the recordsdata listed in step 3 and insert them into the “iceberg_events.”
- When you efficiently add all the information recordsdata, you’ll be able to cease your shoppers from studying/writing to the outdated “occasions” and use the brand new “iceberg_events.”
Some caveats and notes
- In step 2, you’ll be able to management which tables your shoppers/jobs must write to utilizing some flag that may be fetched from exterior sources like surroundings variables, some database (like Redis) pointer, and properties recordsdata, and many others. That manner you solely have to switch your consumer/job code as soon as and don’t need to hold modifying it for every step.
- In step 2, you’re capturing a timestamp that will likely be used to calculate recordsdata wanted for step 3; this may very well be affected by clock drift in your nodes. So that you may wish to sync all of your nodes earlier than you begin the migration course of.
- In case your desk is partitioned by date and time (as most actual world knowledge is partitioned), as in all new knowledge coming will go to a brand new partition on a regular basis, you then may program your shoppers to begin writing to each the tables from a particular date and time. That manner you simply have to fret about including the information from the outdated desk (“occasions”) to the brand new desk (“Iceberg_events”) from that date and time, and you may benefit from your partitioning scheme and skip step 3 completely. That is the method that ought to be used each time potential.
Conclusion
Any massive migration is hard and needs to be thought by means of fastidiously. Fortunately, as mentioned above there are a number of methods at our disposal to do it successfully relying in your use case. If in case you have the power to cease all of your jobs whereas the migration is going on it’s comparatively simple, however if you wish to migrate with minimal to no downtime then that requires some planning and cautious pondering by means of your knowledge structure. You should use a mixture of the above approaches to greatest fit your wants.
To be taught extra:
- For extra on desk migration, please discuss with respective on-line documentations in Cloudera Knowledge Warehouse (CDW) and Cloudera Knowledge Engineering (CDE).
- Watch our webinar Supercharge Your Analytics with Open Knowledge Lakehouse Powered by Apache Iceberg. It features a reside demo recording of Iceberg capabilities.
- Strive Cloudera Knowledge Warehouse (CDW), Cloudera Knowledge Engineering (CDE), and Cloudera Machine Studying (CML) by signing up for a 60 day trial, or check drive CDP. You may also schedule a demo by clicking right here or if you have an interest in chatting about Apache Iceberg in CDP, contact your account staff.
