Analysis of issues in parallel computing, especially parallelized database management. Related subjects include:
The pessimist thinks the glass is half-empty.
The optimist thinks the glass is half-full.
The engineer thinks the glass was poorly designed.
Most of what I wrote in Part 1 of this post was already true 15 years ago. But much gets added in the modern era, considering that:
- Clusters will have node hiccups more often than single nodes will. (Duh.)
- Networks are relatively slow even when uncongested, and furthermore congest unpredictably.
- In many applications, it’s OK to sacrifice even basic-seeming database functionality.
And so there’s been innovation in numerous cluster-related subjects, two of which are:
- Distributed query and update. When a database is distributed among many modes, how does a request access multiple nodes at once?
- Fault-tolerance in long-running jobs.When a job is expected to run on many nodes for a long time, how can it deal with failures or slowdowns, other than through the distressing alternatives:
- Start over from the beginning?
- Keep (a lot of) the whole cluster’s resources tied up, waiting for things to be set right?
Distributed database consistency
When a distributed database lives up to the same consistency standards as a single-node one, distributed query is straightforward. Performance may be an issue, however, which is why we have seen a lot of:
- Analytic RDBMS innovation.
- Short-request applications designed to avoid distributed joins.
- Short-request clustered RDBMS that don’t allow fully-general distributed joins in the first place.
But in workloads with low-latency writes, living up to those standards is hard. The 1980s approach to distributed writing was two-phase commit (2PC), which may be summarized as: Read more
|Categories: Clustering, CouchDB, Data warehousing, Databricks, Spark and BDAS, Facebook, Hadoop, MapReduce, Sybase, Theory and architecture, VoltDB and H-Store||1 Comment|
After visiting California recently, I made a flurry of posts, several of which generated considerable discussion.
- My claim that Spark will replace Hadoop MapReduce got much Twitter attention — including some high-profile endorsements — and also some responses here.
- My MemSQL post led to a vigorous comparison of MemSQL vs. VoltDB.
- My post on hardware and storage spawned a lively discussion of Hadoop hardware pricing; even Cloudera wound up disagreeing with what I reported Cloudera as having said. Sadly, there was less response to the part about the partial (!) end of Moore’s Law.
- My Cloudera/SQL/Impala/Hive apparently was well-balanced, in that it got attacked from multiple sides via Twitter & email. Apparently, I was too hard on Impala, I was too hard on Hive, and I was too hard on boxes full of cardboard file cards as well.
- My post on the Intel/Cloudera deal garnered a comment reminding us Dell had pushed the Intel distro.
- My CitusDB post picked up a few clarifying comments.
Here is a catch-all post to complete the set. Read more
One of my lesser-known clients is Citus Data, a largely Turkish company that is however headquartered in San Francisco. They make CitusDB, which puts a scale-out layer over a collection of fully-functional PostgreSQL nodes, much like Greenplum and Aster Data before it. However, in contrast to those and other Postgres-based analytic MPP (Massively Parallel Processing) DBMS:
- CitusDB does not permanently fork PostgreSQL; Citus Data has committed to always working with the latest PostgreSQL release, or at least with one that’s less than a year old.
- Citus Data never made the “fat head” mistake — if a join can’t be executed directly on the CitusDB data-storing nodes, it can’t be executed in CitusDB at all.
- CitusDB follows the modern best-practice of having many virtual nodes on each physical node. Default size of a virtual node is one gigabyte. Each virtual node is technically its own PostgreSQL table.*
- Citus Data has already introduced an open source column-store option for PostgreSQL, which CitusDB of course exploits.
*One benefit to this strategy, besides the usual elasticity and recovery stuff, is that while PostgreSQL may be single-core for any given query, a CitusDB query can use multiple cores by virtue of hitting multiple PostgreSQL tables on each node.
Citus has thrown a few things against the wall; for example, there are two versions of its product, one which involves HDFS (Hadoop Distributed File System) and one of which doesn’t. But I think Citus’ focus will be scale-out PostgreSQL for at least the medium-term future. Citus does have actual customers, and they weren’t all PostgreSQL users previously. Still, the main hope — at least until the product is more built-out — is that existing PostgreSQL users will find CitusDB easy to adopt, in technology and price alike.
|Categories: Aster Data, Citus Data, Columnar database management, Data warehousing, Database compression, Greenplum, Hadoop, Parallelization, PostgreSQL, SQL/Hadoop integration, Transparent sharding, Workload management||6 Comments|
I stopped by MemSQL last week, and got a range of new or clarified information. For starters:
- Even though MemSQL (the product) was originally designed for OLTP (OnLine Transaction Processing), MemSQL (the company) is now focused on analytic use cases …
- … which was the point of introducing MemSQL’s flash-based columnar option.
- One MemSQL customer has a 100 TB “data warehouse” installation on Amazon.
- Another has “dozens” of terabytes of data spread across 500 machines, which aggregate 36 TB of RAM.
- At customer Shutterstock, 1000s of non-MemSQL nodes are monitored by 4 MemSQL machines.
- A couple of MemSQL’s top references are also Vertica flagship customers; one of course is Zynga.
- MemSQL reports encountering Clustrix and VoltDB in a few competitive situations, but not NuoDB. MemSQL believes that VoltDB is still hampered by its traditional issues — Java, reliance on stored procedures, etc.
On the more technical side: Read more
|Categories: Clustering, Clustrix, Columnar database management, Data warehousing, Database compression, In-memory DBMS, MemSQL, NewSQL, NuoDB, Specific users, Vertica Systems, VoltDB and H-Store, Workload management, Zynga||18 Comments|
Spark is on the rise, to an even greater degree than I thought last month.
- Numerous clients and other companies I talk with have adopted Spark, plan to adopt Spark, or at least think it’s likely they will. In particular:
- A number of analytic-stack companies are joining ClearStory in using Spark. Most of the specifics are confidential, but I hope some will be announced soon.
- MapR has joined Cloudera in supporting Spark, and indeed — unlike Cloudera — is supporting the full Spark stack.
- Mike Olson of Cloudera is on record as predicting that Spark will be the replacement for Hadoop MapReduce. Just about everybody seems to agree, except perhaps for Hortonworks folks betting on the more limited and less mature Tez. Spark’s biggest technical advantages as a general data processing engine are probably:
- The Directed Acyclic Graph processing model. (Any serious MapReduce-replacement contender will probably echo that aspect.)
- A rich set of programming primitives in connection with that model.
- Support also for highly-iterative processing, of the kind found in machine learning.
- Flexible in-memory data structures, namely the RDDs (Resilient Distributed Datasets).
- A clever approach to fault-tolerance.
- Spark is a major contender in streaming.
- There’s some cool machine-learning innovation using Spark.
- Spark 1.0 will drop by mid-May, Apache voters willin’ an’ the creek don’ rise. Publicity will likely ensue, with strong evidence of industry support.*
*Yes, my fingerprints are showing again.
The most official description of what Spark now contains is probably the “Spark ecosystem” diagram from Databricks. However, at the time of this writing it is slightly out of date, as per some email from Databricks CEO Ion Stoica (quoted with permission):
… but if I were to redraw it, SparkSQL will replace Shark, and Shark will eventually become a thin layer above SparkSQL and below BlinkDB.
With this change, all the modules on top of Spark (i.e., SparkStreaming, SparkSQL, GraphX, and MLlib) are part of the Spark distribution. You can think of these modules as libraries that come with Spark.
|Categories: Cloudera, Complex event processing (CEP), Databricks, Spark and BDAS, Hadoop, Hortonworks, MapR, MapReduce, Predictive modeling and advanced analytics, SQL/Hadoop integration, Yahoo||15 Comments|
I frequently am asked questions that boil down to:
- When should one use NoSQL?
- When should one use a new SQL product (NewSQL or otherwise)?
- When should one use a traditional RDBMS (most likely Oracle, DB2, or SQL Server)?
The details vary with context — e.g. sometimes MySQL is a traditional RDBMS and sometimes it is a new kid — but the general class of questions keeps coming. And that’s just for short-request use cases; similar questions for analytic systems arise even more often.
My general answers start:
- Sometimes something isn’t broken, and doesn’t need fixing.
- Sometimes something is broken, and still doesn’t need fixing. Legacy decisions that you now regret may not be worth the trouble to change.
- Sometimes — especially but not only at smaller enterprises — choices are made for you. If you operate on SaaS, plus perhaps some generic web hosting technology, the whole DBMS discussion may be moot.
In particular, migration away from legacy DBMS raises many issues: Read more
|Categories: Columnar database management, Couchbase, HBase, In-memory DBMS, Microsoft and SQL*Server, NewSQL, NoSQL, OLTP, Oracle, Parallelization, SAP AG||17 Comments|
In 1981, Gerry Chichester and Vaughan Merlyn did a user-survey-based report about transaction-oriented fourth-generation languages, the leading application development technology of their day. The report included top-ten lists of important features during the buying cycle and after implementation. The items on each list were very similar — but the order of the items was completely different. And so the report highlighted what I regard as an eternal truth of the enterprise software industry:
What users value in the product-buying process is quite different from what they value once a product is (being) put into use.
Here are some thoughts about how that comes into play today.
Wants outrunning needs
1. For decades, BI tools have been sold in large part via demos of snazzy features the CEO would like to have on his desk. First it was pretty colors; then it was maps; now sometimes it’s “real-time” changing displays. Other BI features, however, are likely to be more important in practice.
2. In general, the need for “real-time” BI data freshness is often exaggerated. If you’re a human being doing a job that’s also often automated at high speed — for example network monitoring or stock trading — there’s a good chance you need fully human real-time BI. Otherwise, how much does a 5-15 minute delay hurt? Even if you’re monitoring website sell-through — are your business volumes really high enough that 5 minutes matters much? eBay answered “yes” to that question many years ago, but few of us work for businesses anywhere near eBay’s scale.
Even so, the want for speed keeps growing stronger.
3. Similarly, some desires for elastic scale-out are excessive. Your website selling koi pond accessories should always run well on a single server. If you diversify your business to the point that that’s not true, you’ll probably rewrite your app by then as well.
4. Some developers want to play with cool new tools. That doesn’t mean those tools are the best choice for the job. In particular, boring old SQL has merits — such as joins! — that shiny NoSQL hasn’t yet replicated.
5. Some developers, on the other hand, want to keep using their old tools, on which they are their employers’ greatest experts. That doesn’t mean those tools are the best choice for the job either.
6. More generally, some enterprises insist on brand labels that add little value but lots of expense. Yes, there are many benefits to vendor consolidation, and you may avoid many headaches if you stick with not-so-cutting-edge technology. But “enterprise-grade” hardware failure rates may not differ enough from “consumer-grade” ones to be worth paying for.
|Categories: Benchmarks and POCs, Business intelligence, Cloud computing, Clustering, Data models and architecture, Data warehousing, NoSQL, Software as a Service (SaaS), Vertica Systems||3 Comments|
The Spark buzz keeps increasing; almost everybody I talk with expects Spark to win big, probably across several use cases.
Disclosure: I’ll soon be in a substantial client relationship with Databricks, hoping to improve their stealth-mode marketing.
The “real-time analytics” gold rush I called out last year continues. A large fraction of the vendors I talk with have some variant of “real-time analytics” as a central message.
Hadapt laid off its sales and marketing folks, and perhaps some engineers as well. In a nutshell, Hadapt’s approach to SQL-on-Hadoop wasn’t selling vs. the many alternatives, and Hadapt is doubling down on poly-structured data*/schema-on-need.
*While Hadapt doesn’t to my knowledge use the term “poly-structured data”, some other vendors do. And so I may start using it more myself, at least when the poly-structured/multi-structured distinction actually seems significant.
WibiData is partnering with DataStax, WibiData is of course pleased to get access to Cassandra’s user base, which gave me the opportunity to ask why they thought Cassandra had beaten HBase in those accounts. The answer was performance and availability, while Cassandra’s traditional lead in geo-distribution wasn’t mentioned at all.
Disclosure: My fingerprints are all over that deal.
In other news, WibiData has had some executive departures as well, but seems to be staying the course on its strategy. I continue to think that WibiData has a really interesting vision about how to do large-data-volume interactive computing, and anybody in that space would do well to talk with them or at least look into the open source projects WibiData sponsors.
I encountered another apparently-popular machine-learning term — bandit model. It seems to be glorified A/B testing, and it seems to be popular. I think the point is that it tries to optimize for just how much you invest in testing unproven (for good or bad) alternatives.
I had an awkward set of interactions with Gooddata, including my longest conversations with them since 2009. Gooddata is in the early days of trying to offer an all-things-to-all-people analytic stack via SaaS (Software as a Service). I gather that Hadoop, Vertica, PostgreSQL (a cheaper Vertica alternative), Spark, Shark (as a faster version of Hive) and Cassandra (under the covers) are all in the mix — but please don’t hold me to those details.
I continue to think that computing is moving to a combination of appliances, clusters, and clouds. That said, I recently bought a new gaming-class computer, and spent many hours gaming on it just yesterday.* I.e., there’s room for general-purpose workstations as well. But otherwise, I’m not hearing anything that contradicts my core point.
*The last beta weekend for The Elder Scrolls Online; I loved Morrowind.
Ever more products try to integrate SQL with Hadoop, and discussions of them seem confused, in line with Monash’s First Law of Commercial Semantics. So let’s draw some distinctions, starting with (and these overlap):
- Are the SQL engine and Hadoop:
- Necessarily on the same cluster?
- Necessarily or at least most naturally on different clusters?
- How, if at all, is Hadoop invoked by the SQL engine? Specifically, what is the role of:
- HDFS (Hadoop Distributed File System)?
- Hadoop MapReduce?
- How, if at all, is the SQL engine invoked by Hadoop?
- If something is called a “connector”, then Hadoop and the SQL engine are most likely on separate clusters. Good features include (but these can partially contradict each other):
- A way of making data transfer maximally parallel.
- Query planning that is smart about when to process on the SQL engine and when to use Hadoop’s native SQL (Hive or otherwise).
- If something is called “SQL-on-Hadoop”, then Hadoop and the SQL engine are or should be on the same cluster, using the same nodes to store and process data. But while that’s a necessary condition, I’d prefer that it not be sufficient.
Let’s go to some examples. Read more
|Categories: Cloudera, Data integration and middleware, EAI, EII, ETL, ELT, ETLT, Hadapt, Hadoop, HBase, Hortonworks, MapReduce, Microsoft and SQL*Server, NewSQL, PostgreSQL, SQL/Hadoop integration, Teradata||36 Comments|
I first wrote about in-memory data management a decade ago. But I long declined to use that term — because there’s almost always a persistence story outside of RAM — and coined “memory-centric” as an alternative. Then I relented 1 1/2 years ago, and defined in-memory DBMS as
DBMS designed under the assumption that substantially all database operations will be performed in RAM (Random Access Memory)
By way of contrast:
Hybrid memory-centric DBMS is our term for a DBMS that has two modes:
- Querying and updating (or loading into) persistent storage.
These definitions, while a bit rough, seem to fit most cases. One awkward exception is Aerospike, which assumes semiconductor memory, but is happy to persist onto flash (just not spinning disk). Another is Kognitio, which is definitely lying when it claims its product was in-memory all along, but may or may not have redesigned its technology over the decades to have become more purely in-memory. (But if they have, what happened to all the previous disk-based users??)
Two other sources of confusion are:
- The broad variety of memory-centric data management approaches.
- The over-enthusiastic marketing of SAP HANA.
With all that said, here’s a little update on in-memory data management and related subjects.
- I maintain my opinion that traditional databases will eventually wind up in RAM.
- At conventional large enterprises — as opposed to for example pure internet companies — production deployments of HANA are probably comparable in number and investment to production deployments of Hadoop. (I’m sorry, but much of my supporting information for that is confidential.)
- Cloudera is emphatically backing Spark. And a key aspect of Spark is that, unlike most of Hadoop, it’s memory-centric.
- It has become common for disk-based DBMS to persist data through a “log-structured” architecture. That’s a whole lot like what you do for persistence in a fundamentally in-memory system.
- I’m also sensing increasing comfort with the strategy of committing writes as soon as they’ve been acknowledged by two or more nodes in RAM.
- I’ve never heard a story about an in-memory DBMS actually losing data. It’s surely happened, but evidently not often.
|Categories: Aerospike, Cloudera, Clustering, Databricks, Spark and BDAS, Hadoop, In-memory DBMS, Kognitio, Market share and customer counts, Memory-centric data management, SAP AG, Theory and architecture||13 Comments|