- SAP HANA is a column-oriented, in-memory database that promises performance gains of up to 100,000x over traditional databases and enables new real-time use cases. Its appliance model reduces costs by simplifying infrastructure requirements. However, it requires new extreme main memory hardware and has limitations for high availability, disaster recovery, and virtualization initially.
- Oracle Exadata is an optimized hardware and software appliance for Oracle Database that scales to hundreds of terabytes. It provides fast performance through SSD caching and compression but does not have a true column-oriented architecture. Additional products like TimesTen and Essbase are needed for optimal OLAP support.
- IBM DB2 with BLU extension provides query acceleration for OL
2. Oracle 11g Release 2
• Self-tuning database components,
• Sophisticated partitioning schemes calling for skilled resources,
• Limited data compression
Oracle 12c
• Oracle 12c is Columnar, compressed, high-speed, in-memory database to directly compete with SAP HANA is what is
promised in the next release of Oracle 12c database.
• Customer will need to purchase, integrate and manage a variety of Oracle products: 12c database, Exadata software and
hardware, TimesTen database, Oracle Clusterware, ASM etc.
HANA – is the best bet and is the comprehensive solution fit for future
• Enables new „real-time“ use cases for customers with up to x100000 performance boost and response times Best
performance guarantee using the appliance model
• Future SAP Business Suite optimized for HANA and all new functionality being built leveraging HANA.
• HANA appliance model reduces operational costs by simplifying classic datacenter layers (no separate efforts needed for
server, operating system, storage, network)
• Less storage needed (data is persistent in Main memory)
Way Forward – Oracle/RDBMS or HANA/In-Memory?
3. Relative Position Of Database Technologies 1/2
Oracle Exadata SAP HANA IBM DB2 BLU
Maturity & age of
technology
Very old, first version developed in late 70’.
Exadata appliance is on the market from 2008.
Running Linux operating system.
Recently launched, 3 years old. Business Suite
on HANA released only in Jan’2013.
Launched in June’2013 as an add on for DB2 V
10.5. BLU is an ‘’Accelerator’’
Number of users /
Sizing
Single appliance (multi node RAC) up to 192 CPU
for SQL processing, 4TB of RAM, 300TB of
storage. Multi-rack infiniband connection makes it
scale up to 18 appliances.
In-Memory technology only limited by size of In-
Memory computing. Single nodes upto 12 TB
available.
No benchmarks available as yet considering its
just come on stage
Applications
supported
Oracle Database 11g/12c.
Accelerates OLAP and OLTP workloads.
HANA DB supports both Business Suite (OLTP)
& BW (OLAP). SAP HANA is optimized for
OLTP+OLAP workloads.
Only works for OLAP workloads. DB2 BLU is
limited to query acceleration on pre-selected
tables solely.
Latency Medium latency, hot/warm data is kept on SSD
storage.
Ultra-low latency as it in-memory DB. Low but Memory is just a caching layer and
query response is still disk bound.
Licensing Licensed by processor core or a Names User Plus. HANA & addl. O/S Licenses. HANA license
options-LREA or Enterprise Edition.
Part of IBM DB2 Advanced Enterprise Edition
Hardware Pre-installed appliance. Comes in 1/8, ¼, ½ and
Full RACK, seamlessly upgradable.
Calls for specific hardware for DB instance. Hardly any installations.
Virtualisation No, each appliance can be cut into smaller ½, ¼
etc.
Only Dev and QA environments. All Environments can be virtualised
Largest Database
Size
Tsted sizes are over 130 TB (PayPal) OLTP
databases. Limited only by storage.
1000TB (= 1PB of data tested) Performance figures available for 10TB of query
data.
Pricing & Delivery
Model
Priced for hardware appliance and software
licensed by CPU or Named User.
Priced per unit (64GB) of in-memory. No direct
relationship with number fo users.
Can be priced per instance & per SAPS
OLAP – Online Analytical Processing e.g. BW, BI, BO Reporting
OLTP – Online Transaction Processing e.g. ERP, CRM, SRM, SCM..
4. Relative Position Of Database Technologies 2/2
Oracle Exadata SAP HANA IBM DB2 BLU
Database model
• Row based on memory/SSD/disk depending
on the data temperature
• Relational
•Column-& Row based in memory
•Relational
•Multidimensional
•Column based in caching
•Pub/Sub
•Object-relational
•XML Database
Integrity model • ACID
•ACID (Atomicity, Consistency, Isolation,
Durability)
•ACID
•Read-Committed
Data Persistence Included in Oracle Exadata Appliance, mixed
SSD/HDD storage.
Depending on the capacity backups can be
done internally or externally.
Part of HANA appliance. External disk for
recovery.
External disk. Data are loaded into main memory.
BW Migration Existing BW running on Oracle Database can be
seamlessly migrated to Exadata
Calls for Dual Stack split, Upgrading Source
System to SAP NetWeaver 7.3 SPS05, Unicode
Compliance before actual migration. DMO–SUM
fast-tracks migration.
Calls for upgrading to DB2 10.5 and then converting
tables from row orgn. to column orgn.
Business Suite
Migration
Application Servers/software can be re-used.
Oracle Exadata comes as an pre-installed
appliance .
Application Servers/software can be re-used.
HANA appliances are required to be set-up.
Relatively speaking-more effort & time is reqd.
to set-up.
Doesn’t support OLTP workloads and isn’t fit for
running Business Suite of for that matter ERP, CRM,
SRM, SCM etc.
Performance Depending on the workload it can be 2x or 200x
times faster.
Atleast 1000 times faster than traditional RDBMS. Published reports suggest- 8 to 25 times faster for
OLAP workloads only.
High Availability &
Disaster Tolerant
Internall RAC and Oracle Data Guard for
replication.
SAP HANA software based replication
or HP Hardware DR with HP Service Guard
Extension.
Leverages IBM’s PureScale solution for HA and DR.
6. Strengths & Weaknesses: Oracle Exadata
Strengths Weaknesses
Oracle Exadata • Innovative hardware model separting storage from
database processing and combining HDD, SSD storage.
• Can scale easly to hundreds of terabytes of data
• Very small risk when migrating from standard Oracle
Database
Alot of experience available.
• Exadata appliance reduces TCO by integrating all classic
datacenter layers (no separate efforts needed for server,
operating system, storage, network)
• No additional storage needed (everything is included in the
appliance)
• Oracle Exadata certified and optimized with SAP
Application from 2011.
• Fast and easy to upgrade online, no downtime, extremely
scaleable
• Strong compression algorythms
• Mature database with several thousands customers
worldwide
• Exadata appliance well tested in many global Top100
companies, PayPal, PG, U.S. Customs, Deutsche Bank)
• SSD Caching technology provides very good performance
gains
• Very mature and reliable in terms of recovery and high
availability
• Certified for OLTP, can support ERP, CRM SRM, SCM,
• Most of the data is kept on traditional SSD/HDD storage
• No virtualization options
• Alot of patch set releases, needs frequent upgrades.
• No column-oriented storage, partly covered by Hybrid Columnar Compression
• Needs Exalytics (Times-Ten, Essbase) to improve the OLAP performance
• Compression algorythms require special handling to be fully effective
• Needs an application layer to do complex calculations
Notes de l'éditeur
A lot of the audience are from industry so this will be of no shock to them but lets just briefly look at a few industry verticals and see how Big Data applies to them in order to get us on the right track of what is required and now what is possible.