Gartner’s ‘Capabilities for Data Management Solutions for Analytics’ Article
What a mouthful ‘Capabilities for Data Management Solutions for Analytics’ is. At least ‘DMSA’ can be used instead. Hang on though – isn’t DMSA an acid used to treat lead, mercury, and arsenic poisoning?!?!?
Anyway, a couple of weeks ago said Gartner article caught my eye on LinkedIn. Team VLDB were at the Teradata Universe event in Nice for most of this week – did anyone else spot Doc Brown from Back to The Future?
The Gartner DMSA article was discussed over beers in Nice, so I thought I’d better follow up on my threat to write about it. Here goes…
What You Need to Know
Tasty morsels in this section include:
- “expectations are now turning to cloud”
- “combining cloud and on-premises in a hybrid configuration is quickly becoming the norm”
- “organisations expect vendors to support them in meeting such deployments”
Despite the above guidance, Gartner then goes on to state that “The ability to deploy on premises, in the cloud or in a hybrid model was not considered a critical capability for this document”.
So, folks want/expect hybrid, but Gartner doesn’t rate hybrid as a critical vendor capability – not yet anyway.
Traditional Data Warehouse Use Case – Categories & Weightings
For the ‘Traditional Data Warehouse Use Case’, Gartner assessed the following capabilities and assigned weightings in order to arrive at overall scores:
- managing large volumes of data = 15%
- loading data continuously = 5%
- query optimisation = 20%
- queries support analytics = 10%
- queries to multiple data sources = 5%
- operational BI use = 5%
- system availability = 10%
- administration/management = 10%
- traditional data warehouse use = 20%
- discovery use = 0%
No doubt we could debate the capability categories endlessly, and argue about the weightings until the cows come home. Let’s not.
For the ‘Traditional Data Warehouse Use Case’, we’d suggest that ‘managing large volumes of data’, ‘query optimisation’ and ‘queries support analytics’ are the most important categories.
Traditional Data Warehouse Use Case – Vendor Scores
Gartner scored the various DMSA technologies as follows, in descending weighted score order:
- Teradata = 4.95
- Oracle = 4.75
- DB2 = 4.13
- SQL Server = 3.90
- Vertica = 3.70
- Greenplum = 3.70
- Hana = 3.55
- Redshift = 3.50
- Snowflake = 3.50
- Insights Platform = 3.40
- BigQuery = 3.38
- MarkLogic = 3.35
- Cloudera = 3.25
- MapR = 3.18
- MemSQL = 3.15
Technologies with a score of 3.0 or less are excluded from the above list as this is the mimimum capability level deemed to meet the use case by Gartner. Those failing to attain a weighted score above 3.0 include MongoDB & Hortonworks. Ouch!
Traditional Data Warehouse Use Case – Analysis
The ‘winner’ according to the weighted scores across all categories, unsurprisingly, is Teradata. They are the world’s biggest pure-play analytics vendor, after all.
However, the *unweighted* scores for what team VLDB would consider the most important categories make interesting reading (green=highest score in category, red=lowest score in category).
Teradata and Oracle both score a perfect 5.0 across the board in the selected categories.
The mega-DBMS vendors Oracle, IBM and Microsoft all score highly with an average in the 4-5 range. We don’t see any of them *ever* in the high-end data warehouse space in which we operate. This is perhaps unsurprising given that in each case their product offering is a general purpose DBMS, rather than an analytics-specific technology.
Vertica and Greenplum attain very respectable scores of 4.0 across the board in the selected categories. Both are built for analytics, so no surprises there.
Hana, Redshift & Bigquery average between 3-4. Hana scores 2.0 for ability to manage large data volumes. Redshift and BigQuery both score 3.0 for query optimisation & analytic query capability. While we can’t comment on Hana, we can confirm that Redshift & BigQuery are off the pace for analytic query optimisation as soon as the query complexity rises much above ‘noob’. Buyer beware etc.
It is interesting to note, for fellow database geeks anyway, that Vertica, Greenplum & Redshift (and Netezza, remember?) all share the same PostgreSQL-derived heritage.
Gartner Vendor Commentary
In addition the the categories, weightings & scores, Gartner also offers vendor commentary. The following comments caught our eye:
BigQuery
- ‘no on-premises version of the product’
- ‘absent or weak functionality was cited by a high number of customers’
Greenplum
- ‘scored well in all four of our defined use cases’
- ‘In the traditional data warehouse use case category, Greenplum scored near the top.’
Teradata
- ‘mature product with market-leading depth and breadth of functionality’
- ‘very high scores for performance’
- ‘strong customer loyalty’
As users of BigQuery, long-term users of Greenplum and *very* long-term users of Teradata, the comments above from Gartner are absolutely aligned with our experiences.
Context Comments
In the section headed ‘Context’, Gartner make the following comments:
- ‘over 80% of the market still adheres to traditional approaches’
- ‘The Hadoop-based group suffered in terms of both execution and vision.’
- ‘Cloud native vendors are seeing increased significance.’
- ‘ease-of-use factor for appliances is matched and exceeded by cloud solutions’.
So, according to Gartner, we want to carry on *largely* doing old-skool data warehousing, the Hadoop ecosystem isn’t ready for this yet, and we’d like to do data warehousing in the cloud, thank you very much!
For those that enjoy a little light reading, the full Gartner article is here.