Analysis of implementations of and issues associated with the parallel programming framework MapReduce. Related subjects include:
I believe in all of the following trends:
- Hadoop is a Big Deal, and here to stay.
- Spark, for most practical purposes, is becoming a big part of Hadoop.
- Most servers will be operated away from user premises, whether via SaaS (Software as a Service), co-location, or “true” cloud computing.
Trickier is the meme that Hadoop is “the new OS”. My thoughts on that start:
- People would like this to be true, although in most cases only as one of several cluster computing platforms.
- Hadoop, when viewed as an operating system, is extremely primitive.
- Even so, the greatest awkwardness I’m seeing when different software shares a Hadoop cluster isn’t actually in scheduling, but rather in data interchange.
There is also a minor issue that if you distribute your Hadoop work among extra nodes you might have to pay a bit more to your Hadoop distro support vendor. Fortunately, the software industry routinely solves more difficult pricing problems than that.
|Categories: Cloud computing, Databricks, Spark and BDAS, Hadoop, MapReduce, MemSQL, Software as a Service (SaaS)||11 Comments|
I’m taking a few weeks defocused from work, as a kind of grandpaternity leave. That said, the venue for my Dances of Infant Calming is a small-but-nice apartment in San Francisco, so a certain amount of thinking about tech industries is inevitable. I even found time last Tuesday to meet or speak with my clients at WibiData, MemSQL, Cloudera, Citus Data, and MongoDB. And thus:
1. I’ve been sloppy in my terminology around “geo-distribution”, in that I don’t always make it easy to distinguish between:
- Storing different parts of a database in different geographies, often for reasons of data privacy regulatory compliance.
- Replicating an entire database into different geographies, often for reasons of latency and/or availability/ disaster recovery,
The latter case can be subdivided further depending on whether multiple copies of the data can accept first writes (aka active-active, multi-master, or multi-active), or whether there’s a clear single master for each part of the database.
What made me think of this was a phone call with MongoDB in which I learned that the limit on number of replicas had been raised from 12 to 50, to support the full-replication/latency-reduction use case.
2. Three years ago I posted about agile (predictive) analytics. One of the points was:
… if you change your offers, prices, ad placement, ad text, ad appearance, call center scripts, or anything else, you immediately gain new information that isn’t well-reflected in your previous models.
Subsequently I’ve been hearing more about predictive experimentation such as bandit testing. WibiData, whose views are influenced by a couple of Very Famous Department Store clients (one of which is Macy’s), thinks experimentation is quite important. And it could be argued that experimentation is one of the simplest and most direct ways to increase the value of your data.
3. I’d further say that a number of developments, trends or possibilities I’m seeing are or could be connected. These include agile and experimental predictive analytics in general, as noted in the previous point, along with: Read more
1. I wish I had some good, practical ideas about how to make a political difference around privacy and surveillance. Nothing else we discuss here is remotely as important. I presumably can contribute an opinion piece to, more or less, the technology publication(s) of my choice; that can have a small bit of impact. But I’d love to do better than that. Ideas, anybody?
2. A few thoughts on cloud, colocation, etc.:
- The economies of scale of colocation-or-cloud over operating your own data center are compelling. Most of the reasons you outsource hardware manufacture to Asia also apply to outsourcing data center operation within the United States. (The one exception I can think of is supply chain.)
- The arguments for cloud specifically over colocation are less persuasive. Colo providers can even match cloud deployments in rapid provisioning and elastic pricing, if they so choose.
- Surely not coincidentally, I am told that Rackspace is deemphasizing cloud, reemphasizing colocation, and making a big deal out of Open Compute. In connection with that, Rackspace has pulled back from its leadership role in OpenStack.
- I’m hearing much more mention of Amazon Redshift than I used to. It seems to have a lot of traction as a simple and low-cost option.
- I’m hearing less about Elastic MapReduce than I used to, although I imagine usage is still large and growing.
- In general, I get the impression that progress is being made in overcoming the inherent difficulties in cloud (and even colo) parallel analytic processing. But it all still seems pretty vague, except for the specific claims being made for traction of Redshift, EMR, and so on.
- Teradata recently told me that in colocation pricing, it is common for floor space to be everything, with power not separately metered. But I don’t think that trend is a big deal, as it is not necessarily permanent.
- Cloud hype is of course still with us.
- Other than the above, I stand by my previous thoughts on appliances, clusters and clouds.
3. As for the analytic DBMS industry: Read more
I spent a day with Teradata in Rancho Bernardo last week. Most of what we discussed is confidential, but I think the non-confidential parts and my general impressions add up to enough for a post.
First, let’s catch up with some personnel gossip. So far as I can tell:
- Scott Gnau runs most of Teradata’s development, product management, and product marketing, the big exception being that …
- … Darryl McDonald run the apps part (Aprimo and so on), and no longer is head of marketing.
- Oliver Ratzesberger runs Teradata’s software development.
- Jeff Carter has returned to his roots and runs the hardware part, in place of Carson Schmidt.
- Aster founders Mayank Bawa and Tasso Argyros have left Teradata (perhaps some earn-out period ended).
- Carson is temporarily running Aster development (in place of Mayank), and has some sort of evangelism role waiting after that.
- With the acquisition of Hadapt, Teradata gets some attention from Dan Abadi. Also, they’re retaining Justin Borgman.
The biggest change in my general impressions about Teradata is that they’re having smart thoughts about the cloud. At least, Oliver is. All details are confidential, and I wouldn’t necessarily expect them to become clear even in October (which once again is the month for Teradata’s user conference). My main concern about all that is whether Teradata’s engineering team can successfully execute on Oliver’s directives. I’m optimistic, but I don’t have a lot of detail to support my good feelings.
In some quick-and-dirty positioning and sales qualification notes, which crystallize what we already knew before:
- The Teradata 1xxx series is focused on cost-per-bit.
- The Teradata 2xxx series is focused on cost-per-query. It is commonly Teradata’s “lead” product, at least for new customers.
- The Teradata 6xxx series is supposed to be able to do “everything”.
- The Teradata Aster “Discovery Analytics” platform is sold mainly to customers who have a specific high-value problem to solve. (Randy Lea gave me a nice round dollar number, but I won’t share it.) I like that approach, as it obviates much of the concern about “Wait — is this strategic for us long-term, given that we also have both Teradata database and Hadoop clusters?”
Also: Read more
|Categories: Aster Data, Data warehouse appliances, Data warehousing, Hadapt, Hadoop, MapReduce, Solid-state memory, Teradata||2 Comments|
My client Teradata bought my (former) clients Revelytix and Hadapt.* Obviously, I’m in confidentiality up to my eyeballs. That said — Teradata truly doesn’t know what it’s going to do with those acquisitions yet. Indeed, the acquisitions are too new for Teradata to have fully reviewed the code and so on, let alone made strategic decisions informed by that review. So while this is just a guess, I conjecture Teradata won’t say anything concrete until at least September, although I do expect some kind of stated direction in time for its October user conference.
*I love my business, but it does have one distressing aspect, namely the combination of subscription pricing and customer churn. When your customers transform really quickly, or even go out of existence, so sometimes does their reliance on you.
I’ve written extensively about Hadapt, but to review:
- The HadoopDB project was started by Dan Abadi and two grad students.
- HadoopDB tied a bunch of PostgreSQL instances together with Hadoop MapReduce. Lab benchmarks suggested it was more performant than the coyly named DBx (where x=2), but not necessarily competitive with top analytic RDBMS.
- Hadapt was formed to commercialize HadoopDB.
- After some fits and starts, Hadapt was a Cambridge-based company. Former Vertica CEO Chris Lynch invested even before he was a VC, and became an active chairman. Not coincidentally, Hadapt had a bunch of Vertica folks.
- Hadapt decided to stick with row-based PostgreSQL, Dan Abadi’s previous columnar enthusiasm notwithstanding. Not coincidentally, Hadapt’s performance never blew anyone away.
- Especially after the announcement of Cloudera Impala, Hadapt’s SQL-on-Hadoop positioning didn’t work out. Indeed, Hadapt laid off most or all of its sales and marketing folks. Hadapt pivoted to emphasize its schema-on-need story.
- Chris Lynch, who generally seems to think that IT vendors are created to be sold, shopped Hadapt aggressively.
As for what Teradata should do with Hadapt: Read more
|Categories: Aster Data, Citus Data, Cloudera, Columnar database management, Data warehousing, Hadapt, Hadoop, MapReduce, Oracle, SQL/Hadoop integration, Teradata||6 Comments|
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
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||14 Comments|
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 talked tonight with Lee Edlefsen, Chief Scientist of Revolution Analytics, and now think I understand Revolution’s parallel R much better than I did before.
There are four primary ways that people try to parallelize predictive modeling:
- They can run the same algorithm on different parts of a dataset on different nodes, then return all the results, and claim they’ve parallelized. This is trivial and not really a solution. It is also the last-ditch fallback position for those who parallelize more seriously.
- They can generate intermediate results from different parts of a dataset on different nodes, then generate and return a single final result. This is what Revolution does.
- They can parallelize the linear algebra that underlies so many algorithms. Netezza and Greenplum tried this, but I don’t think it worked out very well in either case. Lee cited a saying in statistical computing “If you’re using matrices, you’re doing it wrong”; he thinks shortcuts and workarounds are almost always the better way to go.
- They can jack up the speed of inter-node communication, perhaps via MPI (Messaging Passing Interface), so that full parallelization isn’t needed. That’s SAS’ main approach.
One confusing aspect of this discussion is that it could reference several heavily-overlapping but not identical categories of algorithms, including:
- External memory algorithms, which operates on datasets too big to fit in main memory, by — for starters — reading in and working on a part of the data at a time. Lee observes that these are almost always parallelizable.
- What Revolution markets as External Memory Algorithms, which are those external memory algorithms it has gotten around to implementing so far. These are all parallelized. They are also all in the category of …
- … algorithms that can be parallelized by:
- Operating on data in parts.
- Getting intermediate results.
- Combining them in some way for a final result.
- Algorithms of the previous category, where the way of combining them specifically is in the form of summation, such as those discussed in the famous paper Map-Reduce for Machine Learning on Multicore. Not all of Revolution’s current parallel algorithms fall into this group.
To be clear, all Revolution’s parallel algorithms are in Category #2 by definition and Category #3 in practice. However, they aren’t all in Category #4.
|Categories: Greenplum, Hadoop, MapReduce, Netezza, Parallelization, Predictive modeling and advanced analytics, Revolution Analytics, Teradata||Leave a Comment|