Comparison is based on Oracle standard SAGE data warehouse configuration. This includes 14 SAGE cells, 8 DL360 hosts, and Infiniband switches.
Netezza does not have a GbE connection to each Snippet Processing Unit (disk). Instead if has up to 2 GbE connections to each Snippet Processing Array (disk shelf). So the GbE connections are shared among 14 disks.
When a SPU dies, the single SPU that host the mirror for the disk hosted by the dead SPU will start serving requests. In addition, a spare is selected to take over for the dead SPU. The spare regenerates a new primary copy of the data by getting it from the SPUs hosting the mirror. So the SPU hosting the mirror must service requests for the data it owns, plus requests that would have been sent to the dead SPU, plus requests to regenerate the data on the failed mirror. Since the data is (ideally) spread evenly, the SPU hosting the mirror becomes the bottleneck for the entire warehouse.
During disk regeneration after a disk failure, regular processing continues for the bulk of the regeneration. When it’s nearly complete, it causes the system to resynchronize, during which time no new queries can begin. The amount of time in synchronizing can be significant if there are many database-modifying queries in progress during the regeneration. … Recommends avoiding large scale database changes while regenerations are in progress.
What exactly is the disk bandwidth that Netezza claims?
There are four Exadata based hardware products offered by Oracle. They are
HP Exadata Storage Server Hardware SAS 12×300 which has 3.6TB of raw storage, up to 1TB of uncompressed user data, and provides up to 1GB/s of data bandwidth.
HP Exadata Storage Server Hardware SATA 12×300 which has 12TB of raw storage, up to 3.3TB of uncompressed user data, and provides up to 0.75 GB/s of data bandwidth
HP Oracle Database Machine Hardware SAS 300 which is based on SAS-based Exadata Storage Servers. It provides 50TB of raw storage, up to 14TB of uncompressed user data, and up to 14GB/s of data bandwidth.
HP Oracle Database Machine Hardware SATA 1000 which is based on SAS-based Exadata Storage Servers. It provides 168TB of raw storage, up to 46TB of uncompressed user data, and up to 10.5GB/s of data bandwidth.
Comparison is based on Oracle standard SAGE data warehouse configuration. This includes 14 SAGE cells, 8 DL360 hosts, and Infiniband switches.
Netezza does not have a GbE connection to each Snippet Processing Unit (disk). Instead if has up to 2 GbE connections to each Snippet Processing Array (disk shelf). So the GbE connections are shared among 14 disks.
When a SPU dies, the single SPU that host the mirror for the disk hosted by the dead SPU will start serving requests. In addition, a spare is selected to take over for the dead SPU. The spare regenerates a new primary copy of the data by getting it from the SPUs hosting the mirror. So the SPU hosting the mirror must service requests for the data it owns, plus requests that would have been sent to the dead SPU, plus requests to regenerate the data on the failed mirror. Since the data is (ideally) spread evenly, the SPU hosting the mirror becomes the bottleneck for the entire warehouse.
During disk regeneration after a disk failure, regular processing continues for the bulk of the regeneration. When it’s nearly complete, it causes the system to resynchronize, during which time no new queries can begin. The amount of time in synchronizing can be significant if there are many database-modifying queries in progress during the regeneration. … Recommends avoiding large scale database changes while regenerations are in progress.
What exactly is the disk bandwidth that Netezza claims?
One of our Beta customers, a large Retailer, saw significant speedups of their applications while testing Exadata storage. The Exadata based storage system was smaller in size than their existing system and still they saw average speedups of 16X with peak speedups of almost 50X.
Businesses today are increasingly leveraging business intelligence to make important strategic decisions and analyze data to maximize operational efficiency. While the need to process more data increases every day, corporations are also finding their IT budgets being squeezed forcing them to make harder decisions to justify the total cost of ownership (TCO) for their IT software and hardware.
By incorporating Exadata into the IT infrastructure, companies will:
- Accelerate database performance and be able to do much more in the same amount of time.
- Handle growth such that as data and cells are added, I/O bandwidth is also added.
- Deliver mission-critical data availability and protection.
© 2010, www.oracledatabase12g.com. 版权所有.文章允许转载,但必须以链接方式注明源地址,否则追究法律责任.
相关文章 | Related posts:
- Exadata and Database Machine 11.2
- Exadata V2 Sun Oracle Database Machine
- EXTREME PERFORMANCE:Exadata & Database Machine
- I/O Resource Management with Exadata
- Oracle Exadata Overview
- Exadata V2 Pricing
- Extending Automatic Storage Management to Manage ALL Data Oracle Database 11g Release 2
- Storage Portfolio:Oracle is a Storage Company
- Oracle Database 10g Automatic Storage Management Concepts and Administration
- RAC PACK Back-of-the-Envelope Database Storage Design
最新评论