Greenplum has different Nodes

gorgeousvassalSoftware and s/w Development

Nov 7, 2013 (4 years and 2 days ago)

85 views

Greenplum has different Nodes

Loading Nodes
Motion Nodes
Sort Nodes
Join Nodes
Data Nodes
Data Nodes
I
N
T
E
R
C
O
N
N
E
C
T
Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

Perl

Map Reduce


Mini Program

Open Data Flow Processing Engine

SQL

Python

C Functions

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

Web data, Flat files, or remote databases

Data can be internal or external

External Tables

Remote Servers

Example: 450 SQL Server databases seen

as one Greenplum Table

Greenplum allows a table to define data

sources and transports and Transforms data

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

Greenplum Summary

Partnered with SUN to provide Speed
Demon in Data Warehouse Market

Commodity Hardware and Open Source
Postgres make Greenplum affordable

Scale across many processors and servers

Highly Optimized for throughput and streaming

Can use Map Reduce


SQL, Perl, Python, C

Can process data from remote systems

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

5

Dataupia

Founded 2005 by Foster Hinshaw

Father of Data Warehouse Appliance

Co
-
founder of Netezza

I refer to Foster as the “Thomas
Edison of our time”

Coffing Data Warehousing Partner


Foster Hinshaw

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

6

Dataupia Concept

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

7

Dataupia Satori Server

Sa • to •
ri

[
suh
-
tawr
-
ee
]


noun. Key concept in Zen
Buddhism. Refers to deep or

lasting enlightenment.



Ex. With the
DataupiaTM

Satori Server, you can realize
the full potential of your
data.

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

8

Satori Server Specs

Satori uses Linux based Blades.

Blades consist of dual 64
-
bit AMD

Opteron processors.

Each Blade comes with 2 TBs of storage

Each Blade has eight RAID
-
5 Drives

Dataupia Software on the host computer

Captures SQL, parses it, and then executes

It on the Dataupia Satori server.

Dataupia Utilities can extract or replicate

data from existing databases, and while

Loading to Dataupia, create fast indexing.

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

9

Dataupia Satori Server

Integrated bundle of commodity hardware, operating system,

Storage, database interfaces and a dynamic aggregation engine

Bundled into a single blade.

System can be build from 2 to n blades. Each blade

has 2 TB of Disk Space.

Designed to handle a mixed workload of queries

Underlying SQL is all 92 standards and will support all

BI applications from vendors including open source.

System installation takes hours, not weeks, and a

$19,500 entry price is hard to resist.

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

Dataupia Concept

Only Appliance

that doesn’t

make you

change databases

Copyright Coffing Data Warehousing 2008 Phone 937

85
5
-
4838

11

Dataupia Summary

Single Vendor Appliance

Database Transparency

Linear Scalability

Support of many Applications

Mixed Workload

Compellingly low TCO

Foster Hinshaw is a proven star

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

12

HP’s Neoview

Released in 2007

Wal
-
Mart one of their biggest customers

Built with enormous amounts of memory

Challenging Teradata

Coffing Data Warehousing built the Neoview
Courseware

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

13

Neoview Terminology

Node


A single processor and its components


Segment


A series of 16 nodes

System Area Network


Connects all nodes and segments

SQL Engine

Logical Disk

ESAM

Teradata’s PE

Encapsulated SQL

Access Manager is like

Teradata’s AMP

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

14

Neoview Architecture

SQL Engine

ESAM

SQL Engine

SQL Engine

SQL Engine

ESAM

ESAM

ESAM

ESAM

ESAM

Server Net

LDV

LDV

LDV

LDV

LDV

LDV

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

15

Neoview Data Distribution

ESAM

A

B

C

D

ESAM

A

B

C

E

ESAM

A

B

C

D

Neoview has two

Type of tables:


Partitioned



Spread

Across all ESAMs like

Tables A, B, and C.


Non
-
Partitioned



a

Table that resides on

A single ESAM like

Tables D, E, and F.

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

16

Neoview Multi
-
Segment Architecture

Neoview Segment

Neoview Segment

Neoview Segment

Neoview Segment

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

17

Teradata and Neoview Similarities

Both Designed to be Enterprise
Data Warehouses

Both are extremely scalable

Both use Parallel Processing and spread
the data across the processors

Each has a similar architecture and design

Both were led by HP CEO Mark
Hurd

Copyright Coffing Data Warehousing 2008


Phone 937 855
-
4838

18

Teradata and Neoview Differences

Neoview is less expensive

Teradata gives each AMP more disk
space where Neoview gives each ESAM

much more memory

Neoview had the advantage of learning
from Teradata and made improvements

Teradata has more customers and 20 years
experience compared to Neoview

Copyright Coffing Data Warehousing 2008 Phone 937

85
5
-
4838

19

Neoview Summary

HP is a huge name

Mark
Hurd

has excellent leadership and DW focus

Enterprise Data Warehouse Campaign

Going directly after Teradata

HP also owns EDS and Knightsbridge Consulting

Lower costs than Teradata

Design improves on Teradata knowledge