Presto + Hive View Storage/Implementation

I’ve been learning about how presto handles views lately.  This is because we are heavily reliant on presto and we recently ran into multiple use cases where our hive metastore had views which wouldn’t work within presto.

What are Presto Views Exactly?

Presto has its own view implementation which is distinct from a hive’s view implementation.  Presto will not use a hive view, and if you try to  query one, you will get a clear error immediately.

A presto view is based on a Presto SQL query string.  A hive view is based on a a hive query string.  A hive query string is written in HQL (Hive Query  Language), and presto simply does not know that SQL dialect.

How Are Views Stored?

Presto actually stores its views in the exact same location as hive does.  The hive metastore database has a TBLS table which holds every hive table and view.  Views have two columns populated that tables ignore – view_original_text and view_expanded_text.  Hive views will have plain SQL in the view_original_text column whereas presto views will have some encoded representation prefixed with “/* Pesto View…”.   If presto queries a view and does not find it’s “/* Pesto View” prefix, it will consider it a hive view and say that it is not supported.

Making Presto Handle Hive Views

I’ve been doing work for some time to try to make presto-sql support hive views.  I’m using the pull request noted in this issue https://github.com/prestodb/presto/issues/7338 as a template.  It is fairly old though and was made against presto-db rather than presto-sql, so the exercise has turned out to be non-trivial.

I’m still chugging along and will post more when done.  But one thing to note is that this PR does not really make presto support hive views.  It actually allows presto to attempt to run hive views as they are.  Many hive views will be valid presto SQL – e.g. where you’re just selecting from a table with some basic joins and/or where clause filters.

So, this PR basically prevents presto from outright failing when it sees a view that does not start with “/* Presto View”.  It then helps it read the hive query text, possibly lightly modify it, and attempt to run it as if the same had been done for a Presto query.

I plan on doing a number of corrections to the SQL as well; e.g. replacing double quotes, converting back-ticks, replacing obvious function names like NVL with COALESCE, etc.  Eventually I may try to fix more by parsing the hive text with ANTLR or something similar to make as many hive views run by default as possible.  But it will never be a complete solution.  A complete solution would be very hard as it would require a full understanding and conversion of hive language to presto language (which is probably not even possible given some of their differences).

Hive Metastore DB (HMS DB) – Get All Tables/Columns including Partition Keys

How Hive Stores Schemas

I recently had to sync the schemas from a hive database to a normal MySQL database for reasons I won’t bother going into.  The exercise required me to get all columns from all tables in hive for each DB though, and I found that this was not amazingly straight-forward.

The hive metastore DB is a normal MySQL/etc database with a hive schema in it.  The hive schema holds the hive tables though.  So, the information schema is irrelevant to hive; to get the hive table details, you have to interrogate the TBLS table, for example.  To get columns, you need to interrogate COLUMNS_V2, and to get the databases themselves, you look toward the DBS table.

Missing Columns

Other posts I’ve seen would leave you here – but  I found when I joined these three tables together (plus some others you have to route through), I was still missing some columns in certain tables for some reason.  It turns out that partition columns are implicit in hive.  So, in a file system of hive data (like HDFS), a partition column in a table is literally represented by just having the directory named with the partition value; there are no columns with the value in the data.

What this means is that partition columns don’t show up in these normal tables.  You have to look to a separate partition keys table to find them with a separate query.

The Working Query

The query below finds all columns of any kind and sorts them in the order they’ll appear when you select from a table in hive/presto/etc.  I hope it helps you!

select db_name, table_name, column_name from (
SELECT d.NAME as db_name, t.TBL_NAME as table_name, c.COLUMN_NAME as column_name, c.INTEGER_IDX as idx
FROM DBS d
JOIN TBLS t on t.DB_ID = d.DB_ID
JOIN SDS s on t.SD_ID = s.SD_ID
JOIN COLUMNS_V2 c on c.CD_ID = s.CD_ID
WHERE d.NAME = :dbName
union
SELECT d.NAME as db_name, t.TBL_NAME as table_name, k.PKEY_NAME as column_name, 50000 + k.INTEGER_IDX
FROM DBS d
JOIN TBLS t on t.DB_ID = d.DB_ID
join PARTITION_KEYS k on t.TBL_ID = k.TBL_ID
where d.NAME = :dbName
) x
order by db_name, table_name, idx

Hive SQL Standard Authorization – Not Column Level

Hive SQL Standard Authorization Setup

This confluence page https://cwiki.apache.org/confluence/display/Hive/SQL+Standard+Based+Hive+Authorization will walk you through SQL standard authorization in Hive.  The bottom of the page gives advice on setting it up (and it works well on hive 2.3.5 as I just tried it).

As of 6/15/2019, the set up config noted (tested on 2.3.5) is:

Set the following in hive-site.xml:

  • hive.server2.enable.doAs to false.
  • hive.users.in.admin.role to the list of comma-separated users who need to be added to admin role. Note that a user who belongs to the admin role needs to run the “set role” command before getting the privileges of the admin role, as this role is not in current roles by default.
  • Add org.apache.hadoop.hive.ql.security.authorization.MetaStoreAuthzAPIAuthorizerEmbedOnly to hive.security.metastore.authorization.manager. (It takes a comma separated list, so you can add it along with StorageBasedAuthorization parameter, if you want to enable that as well).
    This setting disallows any of the authorization api calls to be invoked in a remote metastore. HiveServer2 can be configured to use embedded metastore, and that will allow it to invoke metastore authorization api. Hive cli and any other remote metastore users would be denied authorization when they try to make authorization api calls. This restricts the authorization api to privileged HiveServer2 process. You should also ensure that the metastore rdbms access is restricted to the metastore server and hiverserver2.
  • hive.security.authorization.manager to org.apache.hadoop.hive.ql.security.authorization.plugin.sqlstd.SQLStdConfOnlyAuthorizerFactory. This will ensure that any table or views created by hive-cli have default privileges granted for the owner.

Set the following in hiveserver2-site.xml:

  • hive.security.authorization.manager=org.apache.hadoop.hive.ql.security.authorization.plugin.sqlstd.SQLStdHiveAuthorizerFactory
  • hive.security.authorization.enabled=true
  • hive.security.authenticator.manager=org.apache.hadoop.hive.ql.security.SessionStateUserAuthenticator
  • hive.metastore.uris=’ ‘

This indeed works well.  After doing this, I was able to set the admin role for my noted user.  This let me create more roles, apply grants on them, and assign them to other users.  Users correctly were rejected from doing queries when they did not have the appropriate grants.

Note that I did not set up any kind of password authentication yet, I was just providing a user name to the hive JDBC connection with no password.  But the grants were properly respected based on the user name.

create role testrole;
grant role testrole to user testuser;
grant select on testdb.sample_data to role testrole;
show grant on table testdb.sample_data;
show current roles;
show principals testrole;

-- Fails for testuser before testdb.sample_data grant is given.
select * from testdb.sample_data;

No Column Level Grants/Restrictions

Unfortunately, I could not find any valid syntax for applying column level permissions, even though columns are shown in the grant outputs.  I also cannot find anything online to imply that permissions can be column level.  So, I think this feature is not supported.  You have to live with table level permissions.

That probably makes sense given there are 2 vendor products in the space for this (Apache Ranger and Apache Sentry).

Presto Doesn’t Work with Apache Ranger (Yet)

Google Group Discovery

After a fairly long fight at building ranger and getting it ready to install, I came across this google group item randomly which made me sad:

https://groups.google.com/forum/m/#!topic/presto-users/gp5tRn9J7kk

It has the following question:

I have setup Presto, Hive, Hue and also setup Ranger for controlling column level access to LDAP users.
Able to see the restrictions getting applied on Hive queries by LDAP users, but however these restrictions are not getting applied on Presto queries.
I understand Presto also uses the same Hive Metastore and Can someone help me why the restricted column access are obeyed in Hive and not Presto when logged in as LDAP user?
And this response:

I am afraid Presto is not integrated with Apache Ranger today. Instead Presto only obeys table-level permissions defined in Hive Metastore.

It’s definitely a roadmap item, we have heard similar requests for integration with Apache Sentry. No specific target date for either at this point.

The Verdict

So, unfortunately, it looks like even if I do finish installing Ranger, I will not be able to get the column level security I’m looking for in Presto.  So, I’m going to move on to analyzing other non-Ranger options.  I’ll also had somewhat ruled out Sentry even before reading this due to a stack-overflow post I read: https://stackoverflow.com/a/56247090/857994 which states:

Just quick update with Cloudera+Hortonworks merge last year. These companies have decided to standardize on Ranger. CDH5 and CDH6 will still use Sentry until CDH product line retires in ~2-3 years. Ranger will be used for Cloudera+Hortonworks’ combined “Unity” platform.

Cloudera were saying to us that Ranger is a more “mature” product. Since Unity hasn’t released yet (as of May 2019), something may come up in the future, but that’s the current direction. If you’re a former Cloudera customer / or CDH user, you would still have to use Apache Sentry. There is a significant overlap between Sentry and Ranger, but if you start fresh, definitely look at Ranger.

I had also already seen numerous other things online agreeing with this and saying that Sentry is weak and Ranger is far more advanced; so this is not surprising.

Eventual Implementation

I found this page https://cwiki.apache.org/confluence/display/RANGER/Presto+Plugin which tells you how to use a ranger-presto plugin.  It was literally made and last edited on May 19th 2019 and refers to version 1.2 of Ranger (the current release).

As I’m writing this on June 9th and 1.2 was released in September 2018 (based on its release note creation date at this site https://cwiki.apache.org/confluence/display/RANGER/Apache+Ranger+1.2.0+-+Release+Notes), this is clearly not released yet.

I double checked on git hub and sure enough, this was just committed 20 days ago.

I wrote one of the committers to get their view on this problem and potential release schedules/etc just for future reference.

Other Options

Apparently Starburst, a Presto vendor company that works on top of various clouds (Azure and AWS), has integrated Sentry and Ranger into their Presto distribution.  You can see that here: https://www.starburstdata.com/technical-blog/presto-security-apache-ranger/.

AWS is also working on Cloud Formation (still in Preview) which supports column level authorization with its Athena (Presto) engine.

 

Hive + Presto + Ranger Version Hell

My Use Case

I was trying to test out Apache Ranger in order to give Presto column-level security over hive data.  Presto itself doesn’t seem to support Ranger yet, though some github entries suggest it will soon.  Ranger can integrate with hive though so that when presto queries hive, the security can work fine (apparently).

Conflicting Versions

I started off by deploying a version of Hive I’ve worked with before; 2.3.5, the latest 2.x version (I avoided 3.x).  After that, I deployed Presto .220, also the latest version.

This was all working great, so I moved on to Ranger.  This is when I found out that the Ranger docs specifically say that it only works with Hive version 1.2.0:

Apache Ranger version 0.5.x is compatible with only the component versions mentioned below

HIVE 1.2.0 https://hive.apache.org/downloads.html

That came from this link: https://cwiki.apache.org/confluence/display/RANGER/Apache+Ranger+0.5.0+Installation.

Alternative Options

I have a fairly stringent need for the security Ranger provides.  So, I was willing to use a 1.x version of hive, depending on what the feature loss was.  After all, quite a few big providers seem to use 1.x.

Unfortunately, the next thing I noticed was that Presto says: “The Hive connector supports Apache Hadoop 2.x and derivative distributions including Cloudera CDH 5 and Hortonworks Data Platform (HDP).”

That is coming from its latest documentation: https://prestodb.github.io/docs/current/connector/hive.html.

I’m not particularly excited to start digging through old versions of Presto as well.

Next Steps

I’m going to try to stick with Hive 2.x for now and a modern version of Presto.  So, my options are:

  1. Research Ranger more and see if it can actually work with Hive 2.x.  Various vendors seem to use Ranger and Hive/Presto together; so I’m curious to see how.  Maybe the documentation on Ranger is just out of date (I know, being hopeful).
  2. Look at Ranger alternatives like Apache Sentry and see if they support Hive 2.x.  Apparently Ranger is beating out Sentry in features, usage, and future support… so I’m not excited about using Sentry.  But if it works, I can always migrate back to Ranger once its support grows for either Hive or Presto.

Update

I starting digging in from JIRA and mailing lists and found that Ranger appears to have had work done on it as early as 2017 for supporting hive 2.3.2.  Here’s a link.  https://issues.apache.org/jira/browse/RANGER-1927.

So, I’m going to give installing ranger a shot on 2.3.5 and see if it works.  If not, I’ll try with 2.3.2 and/or seek community help.  Hopefully I’ll come back and update this afterward with some good news :).