There will naturally be judgments and comparisons made between Oracle’s Exalytics and SAP Hana. This blog is going to present Oracle’s claimed advantages of their Exalytics solution. Keep in mind that the points put forth here are purely from Oracle’s perspective, I will then be producing a follow-up blog that tries to take a more critical look at this comparison. But, for now, let’s take a look at Oracle’s case:.
According to Oracle, TimesTen (their in-memory database) is significantly better than SAP HANA, based on their comaparative strengths over SAP Hana in the following features:
In-Memory Data Caching Times Ten & SAP Hana
In-Memory Columnar Storage Times Ten & SAP Hana
In-Memory Row & Column Compression Only Times Ten fully (SAP Hana has Column only)
In-Memory Indexes Only Times Ten
In-Memory Query Optimizer (Predictability) Only Times Ten
In-Memory NUMA Support (Scale Up) Only Times Ten
In-Memory Parallel Query (Scale Out) Only Times Ten
In-Memory Aggregates & Result Sets Only Times Ten
In-Memory Analytic Functions Only Times Ten
In-Memory Unstructured Data Only Times Ten
High Performance Writes/Updates Only Times Ten
Data Persistence on Disk Times Ten & SAP Hana
Transactional Integrity/Correctness Times Ten; (SAP Hana unknown)
Multi-Version Concurrency Times Ten; (SAP Hana unknown)
Oracle also claims that, compared to Exalytics, SAP Hana has the following limitations:
———-Limited Data Sources with Sybase Replication Server
———-Limited support of 3rd normal form within Business Objects
———-No Parallel Query (Scale-Out) or NUMA (Scale-Up) Support; Limited & Non-Standard SQL
———-Theoretically Possible, but Practically, far too expensive above 2-4 TB
———-Limited Write Performance to update aggregates due to compressed, in- memory columnar storage
Planning & Budgeting
———-Layers of aggregates in SAP BW impact planning on BW on Hana; Limited write performance w/ columnar storage
———-No unstructured data support in Hana; No discovery capabilities across unstructured & structured
Packaged Apps & BI Tools
———-All Packaged Oracle Analytic Applications, Packaged EPM Applications, and any BI Tool works with Exalytics; Hana only works with SAP Tools
In addition to the above, Oracle also says that Exalytics is less expensive than SAP HANA, as follows:
Their claim is that their In-Memory Data Mart, with 512GB compressed data and 1TB memory would cost $825,000, while the equivalent capabilities of an SAP HANA and IBM hardware solution would cost $4,100,000.
Oracle also says that, their package of In-Memory Analytics for Enterprise DW, with 20TB compressed data, and 40TB memory, including 1 Exalytics + 1 Exadata would have a total cost of $2, 500,000.
These same capabilities using SAP HANA and IBM hardware, meanwhile, would require 40 Size L servers to hold all data in-memory, resulting in a total cost of $126,500,000.
Oracle also states that their maintenance and upgrade fees for the above would be substantially lower.
These are the keys points of the Oracle case in this comparison. My follow-up blog to this one will discuss how this viewpoint holds up.