Mitel Fax Memo Manual

							Exhibit 13-l: Module Architecture 
ISA BUS 
Page 49  
						

							13.2.2 Packaging Options 
The Series 6 system is available in four different packaging options: the Model 70, Model 1201, 
Model 12OS, and the Model 640 system. Exhibit 13-2 below outlines each system’s 
specifications. 
Exhibit 13-2: System Specifications 
36.75” (93.34 cm) 
DOC EN60950 
1 IO/220 VAC 
on1 
Parts 15 and 68, 
DOC EN60950 
1 lo/220 VAC 
or 48 
Page 50  
						

							Model 70 
The Model 70 PC-style desktop chassis sits conveniently on a surface and is ideal for small 
areas. This system consists of: 
l An Intel 486 DX66 CPU motherboard with 16 MB system memory, 200 watt power 
supply and a floppy disk drive. 
l Up to two hard disks (IDE interface), which stores system software, prompts, 
messages, mailbox account, and database information for a maximum of 55 hours of 
redundant storage. 
l Up to 7 full slots are available and a maximum 24 ports can be configured. 
The system can be expanded from the basic 10 hours/4 ports by adding pot or by installing 
ports and storage hours. 
Exhibit 13-3: Model 70 
Expansion Slots Offboard Battery Power Supply 
1 
Front Fan and 
Front Bezel 
Motherboard Guides 
Assembly 
Motherboard 
Bay and Carrier for 
Hard Disk and Floppy 
Disk Drive Bay 
(Not Supported) 
51awoc 
Page 5 1  
						

							B. Model 1201 
The Model 1201 system is housed in an upright floor-standing PC-style tower. It consists of: 
l A 486SX33 Plug-in CPU and a backplane that supports up to 11 available slots, with ‘. 
a maximum of 32 ports. 
l Its hard disk interface is IDE and can support up to two hard disks for a maximum 
redundant speech storage capacity of 55 hours. 
l It also includes a floppy disk drive and an internal power supply. 
c. Model 120s 
l The Model 120s consists of a Pentium-100 Plug-in CPU and a backplane that 
supports up to 10 available slots, with a maximum of 60 ports. Its hard disk interface 
is SCSI and can support up to 4 hard disks for a maximum-redundant speech-storage 
capacity of 480 hours. 
l It also includes a floppy disk drive and an internal power supply. 
The system can be expanded by adding ports or by installing both ports and storage. 
Exhibit 13-4: Model 120 
Paver Supgy 
CPU Card 
Line Cards 
Badlane Rope Disk Drive 
Hard Disk Drive 
Page 52  
						

							D. The Model 640 System 
Centigram’s MESA design strategy has evolved to include individual assemblies for each 
component type. The result is the Model 640 (see Exhibit 13-5). The base Model 640 system 
consists of: 
l A storage assembly with one hard disk, one floppy drive 
l A CPU assembly that includes a backplane and up to 15 card slots 
l A power supply assembly with a single power supply. 
The power supply and storage assemblies have their own cooling systems. All assemblies are 
designed for mounting on standard 19” telephony racks. 
r. 
Individual power leads run from the power supply assembly to the CPU, and to a low voltage 
sensor located on the CPU board. A SCSI bus connects the CPU and the hard disks. The floppy 
disk drive is connected to the CPU by a separate standard bus. 
The Model 640 is the optimal configuration for expansion, since the base system uses only a 
small portion of each cabinet. 
l The disk storage assembly can house up to four hard disks and one floppy disk drive, 
giving a maximum storage capacity per module of 1440 hours redundant/2880 hours 
non-redundant. Also housed in the disk storage assembly is a 150 watt power supply 
which supplies power to all of the drives. 
l The CPU assembly holds a single CPU board, a backplane, and can support a 
maximum of 60 ports per module. 
l The power supply assembly holds a 500 watt power supply, which provide enough 
power to service one full CPU module loaded with the maximum number of cards. It 
is capable of monitoring system functions such as fan operations, temperature, and 
power source fluctuations. 
Page 
53  
						

							Power Supply 
Assembly #I2 
Module #2 
Storage 
Assembly #2 
storage 
Assembly #l 
Module #l 
Power Supply 
Assembly #l 
Exhibit 13-5: Model 640 
!I 
i 
Power Supply 
Assembly #4 
Module #4 
Storage 
Assembly #f4 
Storage 
Assembly #3 
Module #I3 
Power Supply 
Assembly #3 
x1937vm6 
Page 54  
						

							.D.I Multihost Operations 
What happens when we take the architecture and expand it across multiple modules? When we 
build a large capacity system, or grow a 60-port system to a 240-port system, all modules operate 
as a single system with a single database, not a virtual single database created by networking 
separate systems together. QNX has built-in peer-to-peer networking, which allows the setup and 
control of tasks on any processor across the network. Whether the tasks are executing on the 
same processor or on physically remote processors does not matter. Centigram’s MESA 
architecture takes advantage of the capability by distributing processing tasks between 
microprocessors on the line cards and on the CPU. In addition, on a multimodule system, an 
entire bus (the MESA-Link bus) is reserved exclusively for CPU-to-CPU message passing. 
The MESA-Link bus acts as a high-speed LAN that transmits all control functions between the 
master and slave modules. This maintains central control over the distributed processing 
functions in a multimodule system. All communications between disk d&s and line cards, 
including database information and speech information, occur over the SCSI bus. The speed of 
the MESA-Link bus, carrying limited control information, ensures that multimodule systems 
perform as efficiently as a single module system. For fault-tolerant needs, a second MESA-Link 
bus is added to ensure that CPU to CPU cormnunication is uninterrupted. 
Ports on a multimodule Model 640 system have equal access to system resources, no matter 
where they are located. Running on the main module processor is a program called “Master,” 
which remembers where all information is located on the hard disk. In a multimodule system, 
this program resides on the primary module. When a call comes in on a line card, the line card 
tells the module, “This is a call for mailbox 569.” Master downloads the information needed to 
process the call onto the line card. If a call comes in on the second module for the same mailbox, 
the second module sends a message to the first module over the MESA-Link bus. Module 1 
sends a copy of the necessary information on mailbox 569 to the second module, also over the 
MESA-Link bus. The second module now knows where all the pointers are, and it can set up a 
data transfer between the line card in the second module and the common hard disk across the 
extended SCSI bus. Master is updated in real-time as changes are made in mailbox parameters 
and/or pointers. 
This dual-bus operation is in contrast to pure networked systems, which do not have the extended 
SCSI bus. The following is an example of a pure networked system: A call comes into the 
Module 1 line card port, but the information needed resides in Module 3. The line card port in 
Module 1 speaks to the CPU in Module 3. The information is retrieved from the hard disk drive 
in Module 3 by its CPU. The information is transmitted back across the local area network to the 
CPU in Module 1. The line card in Module 1 pulls the data from the CPU’s RAM in Module 1. 
Access to a disk comes through a line card in that module or through the CPU. This type of 
design requires multiple database manipulation and support. 
Centigram’s architecture allows the system to work from a single database. Any line card can 
talk directly to any disk in the system to deposit or retrieve digitized speech, without having to 
go through a tier of module processors that either buffer the speech, or control each separate disk. 
Page 
55  
						

							Primary module intervention is required only when information is added to or deleted from a 
hard disk. This makes real-time processing possible even at the highest traffic levels, when all 
ports are active at the same time. 
Continuous System Operation (CSO) software provides a redundant “hot” standby of all primary 
module “master” programs. These programs will reside on the second module and be switched 
over as the “active” master programs for the remaining modules if the primary module fails. The 
hardware component of CSO is the I/O module which monitors the primary module, and should 
a fault be detected, it will move the hardware connections to the back-up module. Hardware 
connections include the console, modem, printer, and a datalink switch integration. 
0.2 Fault Tolerance 
The Model 640 platform has a high degree of fault tolerance when configured with its 
Continuous System Operation (CSO) option, Redundant Drive option and’the Alarm Monitor 
Power Supplies (AMPS). There is NO single point of system failure in the architecture. The 
platform multimodule packaging confines failures to the module level. 
A failure affects only a CPU module. With 4 modules composing a 240-port system, only 25%, 
or 60 ports maximum, will be put out of service. Even multiple CPU module power failures will 
not take out the total system. 
A failure of a disk storage module power supply will put its disk module out of service, but with 
disk redundancy, the redundant disks-residing in other disk storage module-will run the 
system without loss of functionality or system resources. 
Any module has direct access to any disk, including its redundant disk, that resides on a 
redundant bus. The Series 6 platform does not require any functionality in adjacent modules. 
Most other vendors’ systems require messages to go over a system bus between modules to 
access disks that are packaged with line cards. This type of architecture requires at least one other 
system bus and processors in both modules to function in order to pass messages between 
modules. 
The ISA bus is implemented at the CPU module level with up to 15 cards plugged into the 
module ISA bus. There is no systemwide bus that all the cards plug into; therefore, there is no 
bus failure that can disable all the cards. Many competitors have a systemwide bus that can fail 
and disable all cards on that bus, EVEN redundant CPUs. 
The MESA-Link bus is redundant and used for CPU-to-CPU communications. Two 
MESA-Link cards plug into each module. The ISA bus in each module is isolated from the 
systemwide MESA-Link buses. A MESA-Link card failure will not disable the module in which 
the MESA-Link card failed. 
The Model 640 platform software does NOT have a single point of system failure that will 
disable the entire system. The Continuous System Operation (CSO) option will use the 
Page 56  
						

							processors in each module as functionally redundant system processors. There will be a 
STANDBY copy of all System Resourcetasks on a CPU in another module as a backup to the 
CPU of the ACTIVE Primary module. If the Primary module goes out of service, the Continuous 
System Operation software will activate the STANDBY system resource tasks for use by all 
other line card modules within 5 to 40 seconds. In conjunction with the I/O module, resources 
will also move the standby module. 
lb.3 Hard Disk Redundancy 
Since hard disk drives are electro-mechanical devices, they are more prone to failure than circuit 
boards and chips. A drive failure is also the most serious type of system breakdown, since 
irreplaceable stored speech and data are lost, rather than simple functionality for the time it takes 
to replace the part. 
For this reason, Centigram offers full hard disk redundancy across the 
Series 6 platform. 
4 
When redundancy is configured, the system writes to both hard disks, one immediately after 
another, in a technique called “shadow writing.” 
To guarantee integrity between the primary and 
redundant disks, whenever the VoiceMemo application writes to a redundant disk pair, whether 
to store a message or to record a change to the database, both disks must confirm that the task is 
finished before the controlling application registers the task as finished. If one of the disks does 
not report back to the application, the application will retry the operation. If the retry fails again, 
the faulty disk will be put out of service. 
A deactivated hard drive can be removed from the 
drive storage module and a new hard drive inserted while the system continues to process calls. 
“Hot” pluggable hard drives enhance the redundancy feature on the Series 6 Model 640. 
The Series 6 system recognizes the difference between a primary and a redundant disk, but the 
system does not always read from the primary hard disk, even under normal operating 
conditions. Both primary and redundant disks are “mirrored” in that they contain the same 
information. Thus, if a traffic queue starts to build up to a primary drive, the system 
automatically forms a connection to the secondary drive, processes the call, then shadow-writes 
the information back to the primary drive. 
Redundancy is a cost-effective way to provide 
insurance against a fault or a problem in the primary drive _and to improve system throughput in 
high-traffic situations. 
When the VoiceMemo application encounters bad information on a primary disk, the system 
reads the information from the redundant disk immediately. The failure will be unknown to the 
system user, and will require no intervention. The system marks the bad sector on the primary 
disk, and recopies the information from the redundant disk as a background function, without 
interrupting the operation of the system. 
In the event of a catastrophic disk crash, the VoiceMemo application transfers immediately to the 
redundant disk. When the failed disk is replaced, the system begins the shadow writing process 
for all new operations. In addition, the system restores the unduplicated information to the new 
disk as an ongoing background function; full redundancy will be restored within a few hours, 
Page 57  
						

							with no downtime, and no impact on multimode system users. On a single-module system, there 
is a maximum of 15 minutes downtime to activate the new hard drive. 
13.3 Shared Resource Fax 
Beginning with Series 6, FaxMemo resources may be shared across multiple VoiceMemo ports. 
Customers do not need to dedicate a FaxMemo resource to a VoiceMemo port. Instead, a 
FaxMemo resource can be shared by assigning it to multiple VoiceMemo ports. If a call comes 
into one of the ports on the VoiceMemo line group and the caller requests a FaxMemo resource, 
the system “switches” the call over to the FaxMemo resource available to that VoiceMemo line 
group. If none is available, the caller will be prompted that “All fax lines are busy.” To ensure 
the availability of FaxMemo resources, an administrator may dedicate FaxMemo resources to 
VoiceMemo ports in a 1: 1 ratio. For more information on FaxMemo shared resources, please see 
Product Note 2 1. c. 
13.3.1. Shared Resource Fax Configuration Example 
As an example, let’s say a Series 6 Server is going to be used for voice messaging, paging 
notification, and fax mail. Voice messaging requires an incoming VoiceMemo port that listens 
for dial tone, DTMF and line breaks. In this example, we’ll assign two line ports to the 
VoiceMemo inbound line group. Paging notification requires an outgoing port, so we’ll assign 
one port to the paging outbound line group. Fax mail requires an inbound port to receive an 
incoming fax (when callers send faxes into users’ mailboxes). Fax mail also requires an 
outgoing port for users to download their faxes to fax machines. Because Series 6 has FaxMemo 
shared resource capability, however, we don’t need to assign two separate ports to fax mail. 
Instead, we can assign one fax resource to the FaxMemo line group. The resource in this line 
group can be used to perform both incoming and outgoing fax mail functions. How do we do 
this? By assigning this FaxMemo group to the VoiceMemo inbound line group, and also to the 
paging outbound line group. 
It now becomes a fax resource for both incoming functions 
(accepting fax messages and depositing them in users’ mailboxes) and outgoing functions 
(allowing mailbox owners to download their fax messages to fax machines). 
Call Flow for Shared Resource Example (Above): 
A call flow that maps to this server configuration would be the following. An outside caller, 
George, dials a Centigram user’s (Mary’s) telephone number. Mary is not in her office, so the 
call is forwarded to voice mail, and George records a message. This function used one port on 
the VoiceMemo inbound line group. 
Mary has paging capabilities in her mailbox, so she is 
paged when this new message arrives. This function uses one port on the paging outbound line 
group. Mary gets the page, notifying her of a new message, and she dials into her mailbox to 
retrieve that message. This function takes one port on the VoiceMemo inbound line group. 
A few hours later, George dials Mary’s fax telephone number to leave her a fax. The call is 
answered by a port on the VoiceMemo inbound line group. As soon as the port answers the call,