DMX2 Functionality n Components

Copyright © 2006 EMC Corporation. Do not Copy - All Rights Reserved.

DMX-2 1000, DMX-2 2000 and DMX-2 3000 Functionality and Components - 1

© 2006 EMC Corporation. All rights reserved.

DMX-2 1000, DMX-2 2000 and DMX-2 3000 Functionality and ComponentsDMX-2 1000, DMX-2 2000 and DMX-2 3000 Functionality and Components

July, 2006

Welcome to Symmetrix Series Core Curriculum Remote ILT.

These materials may not be copied without EMC's written consent.

EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice.

THE INFORMATION IN THIS PUBLICATION IS PROVIDED “AS IS.” EMC CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Use, copying, and distribution of any EMC software described in this publication requires an applicable software license.

EMC is a registered trademark, Symmetrix, Symmetrix DMX and SymmWin are trademarks of EMC Corporation.

All other trademarks used herein are the property of their respective owners



© 2006 EMC Corporation. All rights reserved. DMX-2 1000, DMX-2 2000 and DMX-2 3000 Functionality and Components - 2

Revision History

CompleteJuly 20061.3

CompleteSept 20051.2

RevisionsCourse DateRev Number

Copyright © 2006 EMC Corporation. All rights reserved.

These materials may not be copied without EMC's written consent.

EMC believes the information in this publication is accurate as of its publication date. The information is subject to change without notice.

THE INFORMATION IN THIS PUBLICATION IS PROVIDED “AS IS.” EMC CORPORATION MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THIS PUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Use, copying, and distribution of any EMC software described in this publication requires an applicable software license.

EMC is a registered trademark, Symmetrix, Symmetrix DMX and SymmWin are trademarks of EMC Corporation.

All other trademarks used herein are the property of their respective owners




DMX-2 1000, DMX-2 2000 and DMX-2 3000 Functionality and ComponentsAfter completing this module, you will be able to:

Locate major components

Identify the card cage directors and memory

Describe how data flows through the DMX-2 1000/2000/3000

Locate and list the function of the ECM and CCM




DMX-2 1000

In this slide you can see that the physical layout of the DMX-2 1000 is very similar to previous Symmetrix models. It has up to four memory cards in the front 12-slot card cage. Disks are both in the front and back of the cabinet. The Service Processor is mounted inside the door. DMX (Direct Matrix Architecture) refers to the improved Director to cache connectivity. DF’s back end Directors and FA’s front end Directors are the same physical board running different Emulation code.




DMX-2 2000

In this slide you can see that the physical layout of the DMX-2 2000 is somewhat similar to previous Symmetrix models. There are up to eight memory cards in the front 24-slot card cage. Disks are both in the front and back of the cabinet. The Service Processor is mounted inside the door. Back-end Directors (DFs) and Front-end Directors (FAs) are now the same physical board running different Emulation code.




DMX-2 3000

In this slide you can see that the physical layout of the DMX-2 3000 is also very similar to previous Symmetrix models. There are up to eight memory cards in the front 24-slot card cage. Disks are both in the front and back of the cabinet. The Service Processor is, as before, mounted inside the door.

The DMX-2 3000 is a three-bay system. It is similar to the DMX-2 2000 but has a third bay containing an additional 288 Disks for a possible total of 576 disk drives in 16 Disk Drive Midplanes (four in Bay 1 front; four in Bay 3 front; four in Bay 1 rear; and four in Bay 3 rear). Each disk midplane contains up to 36 Drives, giving a total of 144 drives in each of the two front bays and 144 drives in each of the rear bays.

Eight Disk Directors are required; the DMX-2 3000 back end consists of (32) 18-Drive loops. Eight slots are reserved for Memory. Eight slots are available for front end host connectivity.




Logical view of the DMX-2 1000/2000/3000Front-end Directors

Back-end DirectorsMemory

Port-Bypass Cards

Disk

Host

The DMX Series models have a similar functional block diagram as can be found in the Symmetrix 8000 Series. Hosts, connected to the back adapters of the front-end directors, send their data to the Symmetrix DMX Series system' s cache. The new Point-to-Point matrix connection between cache and directors allows for high-performance destaging to the drives, or retrieval of data from the disks into cache. Besides the Disk Adapters and drives, the back-end now has a new control card, called a Port Bypass Card (PBC).

The DMX-2 2000/3000 system has eight directors and eight memory boards. The system bandwidth is defined by the number of memory boards, since there are eight memory boards, the bandwidth is 16 GB/s (2GB/s X 8 memory boards). Each director has one connections to each memory board.

The DMX-2 1000 system has eight directors and four memory boards. Since there are four memory boards, the bandwidth is 8 GB/s (2GB/s X 4 memory boards). The architecture has also changed in that each director has two connections to each memory board.




DMX-2 1000 - Slot Configuration (Front)

DIR1

Slot0

BE

DIR2

Slot1

BEor FE

DIR3

Slot2

FE

DIR16

SlotF

BE

DIR13

SlotC

FE

DIR14

SlotD

FE

DIR15

SlotE

BEor FE

M0

Slot

10

M6

Slot

16

M7

Slot

17

M1

Slot

11

DIR4

Slot3

FE

The DMX-2 1000 system has a 12-slot card cage. EMC recommends channel directors be installed inpairs for redundancy. Single board configurations will not require an RPQ. Channel and disk directors are the same (interchangeable) for DMX1000, DMX2000 and DMX3000 but can not be interchanged with DMX-2 directors.

The DMX-2 1000 single bay system has two universal Director slots BE (Backend) or FE (frontend) to enable configurations of Up to six channel directors with two disk directors or an ultra high performance DMX1000-P2 (9 drives per loop) with four disk directors and up to four channel directors. Four slots are reserved for Memory. Dir 1-16 reside in Slot 0-F, Memory boards reside in slot 10,11,16 and 17.

Backend directors are paired for redundancy since the Symmetrix DMX-2 models does not have the dual initiator feature. Director pairing along with dual ported drives and the use of the Port Bypass Cards now provides redundancy for a disk drive failure. Disk director pairing starts from the outside and works toward the center of the card cage. Notice in the diagram above director 1 and 16 are paired and director 2 and 15 are paired. Disk director pairing follows the rule of 17, which states the sum of the directors should equal 17. Front end director pairing configuration are recommended, but not required.




DF/FA DirectorMachine OK

d c b a

d CPU Status

c CPU Status

a CPU Status

b CPU Status

Captive screw

Eject Tab

ResetFRU Blue

LED

Attn Button

Rotary Dials for Diagnostics

Enable Disable Switch

Rotary dial runs from 0-7

‘0’ Represents processor ‘a’, ‘1’ = cpu ‘b’, ‘2’ = cpu ‘c’, ‘3’ = cpu ‘d’

Selections 4-7 are not used

The two-switch set in the middle of the DMX series boards labeled “Rotary Switches for Diagnostics”, run from 0-F and are to be set as prompted during scripts.

The top rotary dial is used for Processor selection and runs from 0-7. 0 Represents processor a, 1 Represents processor b, 2 Represents processor c, 3 Represents processor d, and 4-7 are not used. After choosing the appropriate processor selection, the Attn Button is used to DD a single processor.

The RESET button puts the whole board (all four processors) in DD/IMPL mode.




• Gigabit interfaces• 2GB•1GB•Auto negotiate

• 4 microprocessors (a, b, c, and d)•1GHz Power PC

• 8 ports

• 8 internal links to memory

DF/FA Director

µP d

µP c

µP b

µP a

µP d

µP c

µP b

µP a

µP d

µP c

µP b

µP a

The DMX Series Fibre Channel director has four microprocessors, designated as a, b, c, and d. The four processors control 8 ports. DF/FA boards are identical hardware in DMX-2 1000/2000/3000. Each Fibre Channel director on the Symmetrix DMX Series system supports up to eight internal links to memory. Data transfers between host and cache memory can execute concurrently across all Fibre Channel ports on a director. Fibre is available in single-mode (blue) and multi-mode (black).

Singlemode - The reduced diameter of the core allows all of the light to propagate over the same path. Greater distance can be achieved with single mode than multimode. Requires a LongWave Laser port (LWL) on each end of the cabling.

Multimode - The larger diameter of the core results in multiple propagation paths. Light dispersion limits the distance that can be achieved with multimode. Requires a ShortWave Laser port (SWL) on each end of the cabling.




EA Director

• ESCON Director

• 4 micro-processors• 500 MHz Power PC

• 4 concurrent I/O’s

• 8 ports per board

µP d

µP c

µP b

µP a

µP d

µP c

µP b

µP a

µP d

µP c

µP b

µP a

The EA (ESCON Adapter) has 4 micro-processors (a, b, c & d), with two ports each and is capable of 4 concurrent I/Os. ESCON supports data transfer rates of up to 17 MB/sec with the host. EA boards are available for DMX-2 1000, DMX-2 2000 and DMX-2 3000 models only. Note the port assignments on the back adapter differ from those of the FA (Fibre Adapter).




Multi Protocol Channel Director Adapter DMX-2 1000/2000/3000

µP d

µP c

µP b

µP a

a/A

c/Ad/A

b/A

The Multi-Protocol Channel Director (MPCD) has a 1GHz Power PC microprocessors and combines support for 2 Gb FICON, 1 Gb Ethernet SRDF (GigE), and 1 Gb iSCSI host connectivity on a single channel director. The resulting configuration flexibility reduces costs and maximizes slot efficiency. SRDF is not supported on MPCD FICON ports at this time. SRDF-A is supported on MPCD GigE ports.

Mezzanine card configuration is 1 port per Mezzanine card, 4 Mezzanine cards per director. The DMX-2 1000, DMX-2 2000, and DMX-2 3000 MPCD boards have the Mezzanine boards placed on the back adapter.

FICON requires single mode ports, and iSCSI and GigE require multi-mode ports. The protocol of a port cannot be reconfigured. In the field, combinations of protocols are determined when ordering MPCD and back adapters. Supported combinations of protocols include:

-4 single-mode Mezzanine cards enabling 4 FICON ports.

-4 multi-mode Mezzanine cards enabling 4 iSCSI ports, 4 GigE ports, or

2 iSCSI ports and 2 GigE ports.

-2 single-mode Mezzanine cards and 2 multi-mode Mezzanine cards enabling

2 FICON ports and 2 iSCSI/GigE ports.

-1 single-mode Mezzanine card and 3 multi-mode Mezzanine cards enabling

1 FICON port and 3 iSCSI/GigE ports.




DMX-2 2000 Disk Bay

DMX-2 2000 has 8 Disk cages; 4 in the front and 4 in the rear, each containing up to 36 Drives and 2 Port Bypass Cards (PBCs). Each PBC can access all 36 disks in the Disk cage.




Disk Cage and Midplane

The DMX series Symmetrix incorporates a 2 Gb/s Fibre Channel Arbitrated Loop Back End; this is very different from previous Symmetrix models. Drive mapping in particular has changed substantially. The DMX-2 1000, DMX-2 2000 and DMX-2 3000 are similar with the main difference being the number of DFs (Fibre back end Directors) and the number of Disks. An empty drive cage with the Disk Cage Midplane exposed is shown here with one Port Bypass Card (PCB) in place. Each Disk Cage Midplane contains up to 36 Drives and two PBCs. The drives are dual-ported, connecting to the two PBCs in each cage. The PBC acts as a fibre channel hub as well as detecting drive presence, providing loop control, drive power control, drive reset and drive maintenance control such as the Mark LED function.




DMX-2 1000/2000/3000 PBC

• Communication path to the director is I2C

• Drive presence

• Loop control

• Power control

• Drive reset

• Marker LED

• VPD functions

The Port Bypass Card (PBC) acts as a fibre channel hub as well as performing the following functions: I2C (Inter Integrated Circuit) communication path to the director; Drive presence; Loop control; Power control; Drive Reset; Marker LED and VPD (Vital Product Data) functions.

Star topology protects against loop failure. Each link is separately switched. Signals are regenerated after each link, and signal regeneration and/or retiming is simple to do. Adding devices to the loop does not affect signal quality. In the event of Disk failure or removal, the Drive port is bypassed. The PBC for DMX-2 1000/2000/3000 is similar to the Link Controller Card (LCC) for the DMX801 Disk Array Enclosure (DAE).




Back Adapter

Front Disk MidplaneRear Disk Midplane

PBC

Disk

In order for each processor to access disks in the Front Disk Midplane and Back Disk Midplane, referred to as C and D Devices, it is clear that each processor needs a physical path to two separate Disk Midplane via two cables. The above slide shows the back adapter crossover function. This will allow 16d processor port A and 16c processor port A to access the same Disk Midplane. While 16d processor port B and 16c processor port B access another Disk Midplane. All A ports from processors a, b, c and d access a Front Disk Midplane, all B ports from processors a, b, c and d access a Back Disk Midplane.

If 1cA [Director 1, processor ‘c’, port A] and 1dA are connected to this Disk Cage Midplane, then Dir 16cA and 16dA must also be connected, as failover is processor to processor. In upcoming slides, notice that processors a and b or c and d always share a disk cage. Also note Director A ports are always in the Front Disk Cage Midplane (C Loops) and Director B ports are always in the Back Disk Cage Midplane (D Loops).




Drive AccessDirector 16d

d A

c

b

aBA

BA

BAB

BBAABB

AA

Director 1d

AL_PA=EFDrivePortA

DrivePort

B

16d C0

1d/EF

dA

c

b

aBA

BA

BAB

BBAABB

AA

1d A 16d A

PBC-BPBC-A

The above slide shows how to read the Drive map. In this case, we see this Drive has an Arbitrated Loop Physical Address (AL_PA) of EF. We are looking at a Front Disk Midplane; we know this as Director 1d is using its A port. The failover partner director, Director 16d, through its A port is also connected to a PBC which is connected to the B port on the drive. If a DF/Adapter/PBC fails, then the other drive port will be used to access the disk and the primary port will be bypassed.

This shows how a drive loop is accessed through two different processor ports of pairing directors. Each processor has two ports, each with devices in the Front as well as in the Back Disk Midplane. This is a view looking at the A-ports of director processors 1d and 16d. Director processors' 1d and 16d B-ports have a similar configuration, but in the rear of the machine. With the introduction of the DMX series, directors are paired around Port Bypass Cards (PBCs), which connect to dual ported drives with only 1 port active and the second port bypassed. Failover takes place at the director micro-processor level.

Fibre Channel requires the use of Arbitrated Loop Physical address or AL_PA these are given in hexadecimal. The first or highest priority on the loop is the highest AL_PA EF, although AL_PA’s decrease in increments of 1 some numbers are reserved. This would be confusing to follow so AL_PAs are mapped to a much simpler method of Decimal Loop IDs starting with the first or highest priority on the Loop which is 0. Zero through 126 Loop IDs are supported.




2 Port Active Port Bypass Card (PBC)

The PBC will be configured for two or four ports for director connectivity. In this example, you can see two ports are configured per PBC, allowing two directors (processors) to connect to each PBC. Each PBC connects to the 36 dual-ported drives in a Disk Cage Midplane. A director would be able to access 18 drives through one of these PBC ports. Note that the pairing director would also be able to access the same 18 drives through the other Disk Drive Port via the other PBC, providing an 18-drive loop configuration (two 18-drive loops per Disk Cage Midplane).




4 Port Active Port Bypass Card (PBC)

In this example, you can see four ports are configured per PBC, allowing four directors (processors) to connect to each PBC. Each PBC connects to the 36 dual-ported drives in a Disk Cage Midplane. A director would be able to access 9 drives through one of these PBC ports. Note that the pairing director would also be able to access the same 9 drives through the other Disk Drive Port via the other PBC, providing a 9-drive loop configuration (four 9-drive loops per Disk Cage Midplane).




Back-end Director Pairing 18-drive loopDirector

1d

BB

d A

c

b

a A

BA

B

AB

BB

AA

B

AA

16dC0

1dC1

16dC2

1dC3

16dC4

1dC5

16dC6

1dC7

16dC8

1dC9

16dCA

1dCB

16dCC

1dCD

16dCE

1dCF

16dC10

1dC11

dA

c

b

aBA

BA

B

AB

BB

AA

BB

AA

Director16d

Legend

Primary Connection Director 1d

Bypass Connection Director 1d



PBC

PBC

Each Processor has two ports, each with devices in the Front as well as in the Back Disk Midplane. In the above slide we are showing only one port from Director1d and one port from Director 16d. Notice that each Director has the potential to access all Drives in the loop (18-drive loop configuration in this example). Notice also that using PBC, each director is currently accessing only a portion of the drives (Director 1d has 9 Drives; Director 16d has 9 Drives).

These Directors will have a similar configuration on their second port, which is connected to a different PBC and Disk Midplane. For example, Director 1d has 9 Drives in this Disk Midplane and on its other port it will have 9 drives in another Disk Midplane. Director 16d has 9 Drives in this Disk Midplane, and on its other port it will have 9. Director 1d and Director 16d will be paired in both the front and back Disk Midplanes (only one shown here). With no component failure, each processor will manage 9 drives per port. These reside in Front and Back Disk Midplanes and are referred to as C and D ports. If the processor on Director 1d fails, the processor on Director 16d will now access all 18 Drives on this loop. It will also access all 18 drives on its other port. If the PBC attached to one port on Director 1d fails, the processor port on Director 1d will not have access and Director 16d will have access to all 18 drives on the loop.




Back-end Director Pairing 9-drive loopDirector 1d

dA

c

b

aBA

BA

BAB

BB

AA

BB

AA

BB

d A

c

b

a A

BA

B

AB

BB

AA

B

AA

Director 16d

16dC0

1dC1

16dC2

1dC3

16dC4

1dC5

16dC6

1dC7

16dC8

PBC

PBC

Legend





In the above slide we are showing only one port from Director1d and one port from Director 16d. Notice that each Director has the potential to access all Drives in the loop (9-drive loop configuration in this example). Notice also that using the PBC, each director is currently accessing only a portion of the drives (Director 1d has 4 Drives; Director 16d has 5 Drives).

These Directors will have a similar configuration on their second port, which is connected to a different PBC and Disk Midplane. For example, Director 1d has 4 Drives in this Disk Midplane and on its other port it will have 5. Director 16d has 5 Drives in this Disk Midplane, and on its other port it will have 4. Director 1d and Director 16d will be paired in both the front and back Disk Midplanes (only one shown here). With no component failure, each processor will manage 4 drives on one port and 5 Drives on the other. These reside in Front and Back Disk Midplanes and are referred to as C and D ports. If the processor on Director 1d fails, the processor on Director 16d will now access all 9 Drives on this loop. It will also access all 9 drives on its other port. If the PBC attached to one port on Director 1d fails, the processor port on Director 1d will not have access and Director 16d will have access to all 9 drives on the loop.

Directors are paired Processor to Processor using the rule of 17. This means mirrors will NOT be placed across Directors using 17 rule (unless only 2 Directors are present). Directors providing redundant paths to Disks (Director 1d and Director 16d above ) will not use the same PBC in order to maintain redundancy on the PBC level. The PBC acts as the hub for all the Fibre disk drives in the disk cage. The PBC contains the switch elements and control functions to allow intelligent management of the two FC-AL loops embedded in each disk cage midplane. There are two PBC’s per disk cage midplane.




Disk Map 18-Drive Loop

Pairing disk directors are able to access each others drives in the event of a disk director failure. The Disk Map displays the director pairs one above the other and the microprocessors have the same physical position on the director, for example 1a and 16a, 1b and 16b etc.. This is a view of the Disk Map in Symmwin showing a 18 drive per loop or Performance configuration. Green shading is used to indicate Director primary drive, notice each director has 9 physical devices on one Loop and 9 on the other. 1a for example has 9 C devices C0, C2, C4…. (A port also front of Symmetrix) and 9 D devices D1, D3, D5…..etc.

Notice that in the first column of drives, every other drive is accessed by one of the two processors associated with that loop. For example, we have drive 1aC0, 16aC1, 1aC2, 16aC3, etc. This “every other” drive access pattern is new to the DMX series machines because of the Fibre Channel Arbitrated Loop technology used in the DMX series Symmetrix machines, versus the SCSI technology used in Legacy model Symmetrix machines.




Disk Map 9-Drive Loop

This is a view of the disk map in Symmwin showing a 9 drive per loop or Performance configuration. Green or white shading is used to indicate the director’s primary drives, Notice each director has five physical devices on one Loop and four on the other. For example, 1a has 5 C devices (director A port also front of Symmetrix) and 4 D devices (director B port also rear of Symmetrix). Pairing directors are able to connect to all 9 drives in the event of a director or PBC failure. DA-1aC0 is selected. Also highlighted is the M2 device DA-15bD0. The M1 is in a front Disk Midplane and the M2 is in a rear Disk Midplane.




DMX-2 2000/3000 - Slot Configuration (Front)

DIR1

Slot0

BE

DIR2

Slot1

BE

DIR8

Slot7

FE

DIR3

Slot2

FE

DIR4

Slot3

FE

DIR5

Slot4

BEor

FE*

DIR6

Slot5

BEor

FE*

DIR7

Slot6

FE

DIR9

Slot8

FE

DIR10

Slot9

FE

DIR16

SlotF

BE

DIR11

SlotA

BEor

FE*

DIR12

SlotB

BEor

FE*

DIR13

SlotC

FE

DIR14

SlotD

FE

DIR15

SlotE

BE

M2

Slot

12

M0

Slot

10

M4

Slot

14

M5

Slot

15

M6

Slot

16

M7

Slot

17

M3

Slot

13

M1

Slot

11

*DMX3000 requires 8 backend dirs

The DMX-2 2000 and DMX-2 3000 have the same 24-slot card cage. EMC recommends channel directors be installed in pairs for redundancy.

The DMX2000-M2 two-bay system has four universal Director slots. The universal slots are used for either Back-end (BE) or Front-end (FE) to enable configurations with four disk directors and up to twelve channel directors for the DMX2000-M2 (18 drives per loop) or eight disk directors and up to eight channel directors for performance DMX2000-P2 (9 drives per loop). Eight slots are reserved for Memory. Backend director pairing along with dual ported drives and the use of the Port Bypass Cards now provides redundancy for disk drive path failures while RAID and hot spares provide protection against disk failures.

The DMX3000-M2 three-bay system has eight disk directors and up to eight channel directors in an 18-drive loop configuration only. Eight slots are reserved for Memory.




DMX-2 2000 & DMX-2 3000 Midplane Rear

ECM

CCM

The DMX-2 2000 and DMX-2 3000 systems have a 24-slot midplane. On the rear side of the midplane, the adapters reside behind their respective directors, and four slots behind the memory are reserved for the CCM and ECM pairs. The remaining four slots are for the power bus bars.

Cable routing is via the top of the cabinet, as well as the bottom.




Direct Matrix Architecture

The DMX Series system currently uses M5 memory boards. Each memory board has sixteen ports, one to each director. Each region can sustain a data rate of 500MBs, 4 regions per card, so 2GB per card. If a director is removed from a system, the usable bandwidth is not reduced. If a memory board is removed the usable bandwidth is dropped by 2GB/s. Each memory board has 4 memory regions, each region has a 500MB bandwidth. If more than 2 x 250 MB paths attempt to access the same memory region then round-robin arbitration is implemented (they must share the 500MB bandwidth).




Communications and Environmentals

Director-to-Director Communications will bypass the data path via the Communications Control Module (CCM) using the I2C Bus. The CCM is a high-performance network interface connecting to each of the CPUs on the director boards. CCM carries control traffic only, bypassing memory. Environmental monitoring is a function of the Environmental Control Module (ECM).




Functions

The Environmental Control Module (ECM) monitors and logs environmental events for all DMX Series field replaceable units (FRUs) and reports any operational problems, such as thermal excursion, voltage droop, etc. FRUs include devices such as director boards, memory boards, port bypass cards, power supplies, fans and switches. The ECM also centralizes system FRU management by acquiring all vital product data (VPD), status and control of FRUs via the I2C bus or an Ethernet interface. The ECM remotely monitors temperature, voltage and fan speed remotely.

The Communications Control Module (CCM) handles all maintenance and control of the communications functions in the DMX Series system. It acts as a communications agent between the service processors and the director processors, provide capabilities for diagnostics and monitoring of network interface, and provide local CCM hardware control functions. Redundant CCMs are required to prevent a CCM component from becoming a single point of failure.




DMX-2 1000/2000/3000 ECM CCM Matrix

Both ECM0 and ECM1 communicate with CCM0, CCM1, fans, power supplies, memory boards, door switch, PLIM0, PLIM1 and both batteries. ECM0 communicates with directors 1 to 8, half of the PBCs, and ECM1. ECM1 communicates with directors 9 to 16, the remaining PBCs, and ECM0.




Comparison DMX-2 1000 vs. DMX-2 2000/3000

16GB/s16GB/s8GB/sSystem Bandwidth

2 (N+1)2 (N+1)N+1Power Redundancy

Three PhaseSingle/Three Phase

Single PhaseInput Power8 4-82-4Disk Director #

193-57696-28848-144Min-Max DisksFC-ALFC-ALFC-ALDisk Type

Point-to-PointPoint-to-PointPoint-to-PointCache Interconnects

8 (0-7)8 (0-7)4 (0-1) (6-7)Memory Slots

24 Slots24 Slots12 SlotsCard Cage Slots

3-Bay2-Bay1-BayCabinetDMX-2 3000DMX-2 2000DMX-2 1000

The physical attributes of the DMX-2 1000 are almost identical to that of the DMX-2 2000/3000 with a few exceptions: 12 slots vs. 24 slots, System aggregate bandwidth of 8 GB/s versus 16 GB/s due to 4 versus 8 memory boards, Power redundancy N+1, not 2(N+1).

Minimum Drive count shown here may vary, always check the latest configurations supported.




Module Summary

Key Points covered in this module:

Major components Location

Card cage directors and memory locations

Data flow through the DMX-2 1000/2000/3000

ECM and CCM location and function




Closing Slide

DMX2 Functionality n Components

Documents

applicable software license

direct matrix architecture

vital product data

slot card cage

disk cage midplane

fibre channel hub

back disk midplanes

environmental control module