They say you cant be too careful. This adage also holds true in the
telecom arena, especially as regards your PBX, auto-attendant, IVR, and voice mail. It is
difficult to estimate how many companies run their telecommunication systems without
proper fail-over measures, but we are willing to bet that many do.
That so many of us should live dangerously is odd, since most leave
nothing to chance when we buy our telecommunication systems. We do all kinds of research,
basing our decisions on MTBF (Mean Time Between Failure) numbers, the manufacturers
reputation for quality, and our gut feeling, among other things. But once the system is
bought and installed, it ends up in a closet, largely forgotten, except for occasional
maintenance.
The fact is, systems fail. And, in computer systems, the hard drive is
often the first component to go. Some people take comfort in the fact that they perform
nightly backups, but if you are in a business that requires 100percent up-time from your
telephone system, a hard-disk crash could take precious hours to restore.
Enter Clariions TeleStor, a fully redundant disk system targeted to
the CTI and telecom industry. In its heart, the TeleStor disk array offers the power and
availability of RAID [Redundant Array of Independent (or Inexpensive) Disks] which is
todays weapon of choice against disk failures.
Clariion (a Data General company) has long been in the RAID business and
now with TeleStor it has made a foray into the CTI industry as storage, high availability,
and data integrity requirements have been on a marked increase.
INSTALLATION
There are two steps in deploying TeleStor the hardware and software portions. The
hardware consists of the chassis (containing the disk drives, storage processors, and
fans), an external power supply, and SCSI adapters. The software, which is used to manage
and monitor the disk array comes in two flavors. The server can either be set up to
establish a direct port interface to TeleStor through the DiskArray Storage System Manager
(DASSMGR) utility or the ArrayGUIde utility can be installed on the server. ArrayGUIde has
a neater GUI interface and can utilize the power of client/server computing to access
TeleStor.
Preliminaries
For our installation, we used a Pentium 200-MMX PC with 32 Megabytes of RAM
running Windows NT Server 4.0. We installed the recommended BusLogic MultiMaster SCSI-2
card and after configuring the card we were ready to interface the PC to the disk array.
Our TeleStor series 1900T had 2 rows of 5 disks each for the total of 10 hard drives. Each
drive had a capacity of 8.8 GB giving us a total of 88 GB of storage. The actual capacity,
however, would normally be slightly less, as RAID requires a moderate amount of overhead
for its operation.
The system was served by 2 storage processors (SPs) and included two fan modules and
two power supplies (DC) as well as a battery backup unit (BBU). The SPs work cooperatively
to pass data between the server and the disk array at maximum bandwidth. In the event of a
power outage, the BBU provides power to the chassis long enough for the cache to be
flushed and the disk array to be powered down in a normal state. The storage processors
are outfitted with SCSI-2 adapters which we used to interface to our server.
Server Arrangements
TeleStor can be connected to a single server or to dual servers. For the single server
arrangement, TeleStor can be configured in one of three ways. The basic configuration
consists of one storage processor connected to one SCSI-2 adapter on the server. In the
dualSP configuration, TeleStor is outfitted with 2 storage processors which are daisy
chained to a single SCSI adapter on the server. For the dual-adapter/dualSP configuration,
the server has 2 SCSI-2 cards, which are separately connected to 2 storage processors on
TeleStor.
For the dual server arrangement, TeleStor can also be configured in one of three ways.
The split-bus configuration consists of connecting one SCSI-2 adapter from each server to
the two storage processors on TeleStor. The dual-initiator configuration is similar to the
split-bus configuration except that the storage processors are daisy chained, making a
continuous line between the SCSI-2 adapters of the 2 servers. Finally, for the
dual-initiator/ dual-bus configuration each server is outfitted with 2 SCSI-2 adapters,
then each storage processor is connected to one adapter from each server.
TeleStor Configurations
The six different TeleStor configurations offer varying amounts of reliability
and data availability, starting with the basic configuration, which offers the lowest
reliability to the dualinitiator/ dualbus configuration, which offers the highest. The
fact that TeleStor offers so many options makes it a strong product as there are
sufficient number of configurations to satisfy any requirement. Within TeleStor, disks are
arranged as LUNs. A LUN (Logical Unit Number) consists of one or more disks under the same
ID number viewed by the operating system as one drive. For example, Windows NT would
assign a different drive letter to each LUN presented by TeleStor. After determining the
participating disks for a LUN, the disks must be binded. This creates the logical unit
recognizable by the operating system.
We set up our system with the dualSP configuration, which entailed daisy chaining the
storage processors and interfacing one of them to the SCSI-2 adapter on our server. Next,
we connected TeleStor to an external power supply (the system does not come with its own
AC power supply) and we switched the system on.
ArrayGUIde Software Installation
While the drives were spinning up, we started the ArrayGUIde software
installation. We decided not to use the DASSMGR software as the ArrayGUIde utility
provides for a much better interface.
ArrayGUIde comes in many flavors in order to support the variety of operating systems
in the market. The supported operating systems include the popular flavors of UNIX, Sun
Solaris, Novell NetWare, and Windows NT. ArrayGUIde is actually composed of two
components, the Agent and the Manager. The Agent component runs on the server and directly
communicates to TeleStor over the SCSI-2. It acts as the liaison for the ArrayGUIde
Manager component which provides the user interface. With the exception of NetWare, which
does not have a GUI, the ArrayGUIde Manager utility can be installed on the server or on
clients (including Windows 95), which have network access to the server.
Agent Installation: We started by copying files into the Clariion subdirectory on the
servers local drive. We then used the Configuration Editor screen (Figure 16) to
configure the Agent to communicate with the storage processors. Here, one can set the
polling rate of the Agent, contact information, and users authorized to configure TeleStor
through the Manager. Since the Agent runs as a service under Windows NT and uses Remote
Procedure Calls to communicate to the Manager, we made sure that all these services are
set for automatic startup at boot time using the Services applet of the Control Panel.
Manager Installation: The ArrayGUIde Manager, which we decided to set up on our server,
is a friendly, graphicalbased application used to administer and monitor TeleStor. The
Manager can be set up to access all the TeleStor chassis connected to the server as well
as multiple servers on the network. This approach allows one host or multiple hosts on the
network to access all of the TeleStor chassis, thereby facilitating the management and
monitoring of all the disk arrays on the network.
We launched the ArrayGUIde Manager program, which prompted us with screens to select
the servers and then select the chassis that we wanted to connect to. In our case, there
was only one host and one chassis which we selected from the screens. ArrayGUIde Manager
then presented us with its main screen which was a graphical representation of the actual
TeleStor chassis (Figure 17). This was the Equipment View under the Views menu item.
Confirmation
Our TeleStor disk array had been previously set up with two RAID 5 LUNs, each consisting
of 5 disks. We verified this by choosing the Summary View from the Views menu item. This
action launched a new screen show-ing the drive components of the disk array in more
detail (Figure 18). From this screen, we verified that there were 2 LUNs configured as
RAID 5 on our system. We could also see which disks belonged to which LUNs as well as
which storage processor was serving a particular LUN to our server. One of the great
features of the Equipment or the Summary Views was the ability to instantly get
information about any part of the disk array just by clicking on that part.
For example, when we clicked on one of the disks, we were presented with a screen
displaying information about the disk (Figure 19). We were also able to get information
for the storage processors and the LUNs in the same way. One thing ArrayGUIde Manager is
not short of is the informational and monitoring screens. We were able to get a good
amount of information on the SPs and the LUNs by choosing the View Settings item from the
Options menu. This action popped up a screen filled with information about the SPs and
LUNs (Figure 20).
The Equipment View also allows one to get information on the fans, internal power
supplies, and the battery backup unit by clicking on the areas where these systems are
located.
Color Coding
The components of TeleStor are displayed using color coding to signify their states. For
example, amber signifies a faulted state, while green signifies a normal state. The Legend
menu item contains 2 sub-menus to show Disk Legend and LUN Legend. Choosing either of
these items brings up a screen displaying all possible color codes together with their
meanings. At first we thought that having a Legend menu item on the main screen was a bit
odd, but nevertheless, it could prove to be useful when needed.
We must say that although some of the ArrayGUIde Managers GUI screens do not
quite conform to the Windows model (e.g., no Help menu item, unusual screen colors, etc.)
we found the program very userfriendly and a pleasure to use.
DOCUMENTATION
We were very impressed with the printed documentation for TeleStor. There were several
manuals delivered with the system that covered everything one needed to set up and
maintain the disk array. The manuals were:
Installing And Maintaining A CLARiiON TeleStor Series 1900T DiskArray Storage
System This manual covers the entire hardware installation of TeleStor as well as
upgrading and repairing the system.
- Using The Disk-Array Storage System Manager (DASSMGR) To Configure And Manage A CLARiiON
Disk-Array Storage System This manual covers the DASSMGR application which runs
over a serial connection the disk array and is used to administer and monitor TeleStor.
- Setting Up A Server For A CLARiiON Disk-Array Storage System This manual
thoroughly covers the setup and configuration of several supported operating systems
including multiple versions of NetWare, SCO UNIX, and Windows NT servers. The manual also
has a section on installing and configuring the SCSI-2 adapter in the server.
- Using The ArrayGUIde Utility Revision 1.6 To Configure And Manage CLARiiON Disk-Array
Storage System
This manual covers the ArrayGUIde Agent and Manager for several supported
operating systems as well as providing comprehensive instructions on using ArrayGUIde
Manager.
As much as we liked the TeleStor manuals we were unimpressed with the online help for
Array GUIde. Array GUIde Manager online help needs some improvements to make it easier to
locate information, more comprehensive, and better-enabled for context sensitive help.
FEATURES
The TeleStor disk array provides the RAID storage solution for demanding Central Office
(CO) and CTI applications. It supports up to 10 disk drives for a total of 88 GB of data
storage. For larger installations, multiple arrays can be attached to a single SCSI
channel. Some of TeleStors features are:
- Full modem RS-232 connection permits integration with existing facilities management
systems.
- Redundant subsystems and power feeds eliminate single points of failure.
- Global sparing provides automatic, unattended disk rebuild, reducing vulnerability to a
second drive failure.
- Full data access and integrity is assured in the event of a power supply, storage
processor, disk drive or fan failure.
- User installable, configurable, and maintainable design increases productivity.
- Vendor-independent architecture accommodates a variety of host servers.
ArrayGUIde Agent and Manager are software components providing the user with a GUI
environment to administer and monitor TeleStor. Some features of the ArrayGUIde are as
follows:
- Supports multiple operating systems.
- Provides redundant, active SCSI I/O paths for one or more servers
connected to one or more CLARiiON disk arrays.
- Eliminates SCSI bus adapter, SCSI cable, and storage processor as single
points of failure.
- Automatically executes I/O path rerouting in seconds when triggered by
SCSI errors/time-outs or SCSI adapter failure.
- CLARalert, which is an add-on service program delivering early
notification (e.g., pager) to CLARiiON customers of service events at their customer
enduser sites can be deployed with TeleStor. CLARalert also enables service organizations
to remotely dial-in to enduser sites to perform service diagnostics. CLARalert relies on
ArrayGUIde Agent for event notification.
OPERATIONAL TESTING
Now that we had a good familiarity with the ArrayGUIde Manager program, it was time to put
the system to the test.
Reading/Writing Data
We started by writing and reading large blocks of data to and from the two RAID 5 LUNs on
the disk array. TeleStor performed almost on par with other systems with regular SCSI-2
disks, at times edging them out on performance. Of course, having two storage processors
available on our system was, in itself, a good performance factor, as they work in
aggregate to provide maximum bandwidth.
Simulated Hard Disk Crash
Next, we simulated a hard disk crash by pulling one of the hard drives out of the
chassis during a read operation. This action brought about a small pause in accessing the
LUN as TeleStor began to deal with this event. Within moments, the read operation
continued as if nothing had happened. The same scenario was repeated during a write
operation. ArrayGUIde Agent was right on top of things as the simulated disk crash
presented us with an alert window declaring a cabinet fault (Figure 21). The Equipment and
the Summary views of the ArrayGUIde Manager also indicated the fault by colorizing the
failed disk drive with amber. This exercise certainly made it clear to us the importance
of having RAID protect the system against disk crashes. Had this been a missioncritical
system, a real disaster would have been averted as the system would simply ride through
the disk crash.
Polling
The ArrayGUIde Agent and Manager can be configured with various polling intervals
to inspect the TeleStor status. The polling feature is accessed using the Options/Poll
Settings menu item allowing the user to set the polling rate to several values or simply
disable it. For ArrayGUIde Manager, the user can also use the Poll button from the
Equipment or Summary View screens to manually poll the disk array.
Replacing the crashed hard drive was as easy as pulling out the bad drive and pushing
in a new one all the while the system kept running. At this point, the drive color
was changed to blue on the ArrayGUIde Manager screen indicating a rebuild in progress as
the newly inserted disk was being integrated into the RAID LUN. We wondered what would
happen if two disks in the same LUN crashed simultaneously so we pulled out 2 drives
during a read operation.
This caused an immediate halt on disk access, but once we re-inserted the drives,
TeleStor recovered, and it was back in business. Statistically speaking, chances of two
drives within the same LUN crashing simultaneously is minuscule, but we think TeleStor
reacted just fine to this condition by halting its operation rather than crashing.
Reconfiguring
We decided to make some changes to TeleStor by creating our own disk
configuration. In order to do this, we needed to free up a few disks, so we decided to
unbind one of the RAID 5 LUNs. We did this by going to the Summary View screen and
clicking on the particular RAID 5 LUN and choosing the Unbind command (Figure 22). There
is no way to unbind drives from a menu command, which could be cause for some confusion.
We believe that all the functionality of a program should be accessible through the menu
as well, so this area of ArrayGUIde Manager would be a good candidate for improvement.
Unbinding the LUN was a straightforward task and it was preceded by a couple of
confirmation screens warning us of total data loss on the LUN. After the unbind was
completed, we used the Bind button to get to the bind screen (Figure 23). From this
screen, we were able to create 2 LUNs, one configured as RAID 1 and the other as a
standalone drive. We liked the fact that the bind screen indicated the unbinded disks by
placing a checkbox above them. This way, we could easily select from the available disk
pool to participate in the new LUN by clicking on the corresponding checkboxes. One
problem we ran into was trying to partition and format one of the newly created LUNs
before it had finished binding (the binding process could take a few hours depending on
the RAID level) which made the LUN inaccessible.
We called Clariion technical support and, sure enough, the problem was our impatience
to partition the LUN before it had finished binding. Once we had cleared that problem, the
rest was smooth sailing. After the LUNs were binded, we used the Disk Administrator
program from Windows NT to partition and format the LUNs with the NTFS file system. The
drive letters corre sponding to the LUNs were immediately created and were available for
usage (for once we didnt have to reboot Windows for something to take effect) as
evident from Windows Explorer.
Application Transparent FailOver
An add-on software product, called Application Transparent FailOver (ATF), is used to
maximize the investment made in the redundant SCSI-2 adapters and storage processors for
TeleStor by automatically shifting the load from a failed module to the other one. Without
ATF, one would need to manually reconfigure the system when a SCSI adapter or storage
processor failure occurs. We tested ATF by pulling out one of the storage processors while
TeleStor was running. This action caused ATF to automatically place all the available LUNs
under the other storage processor as we could see from the summary view. Once we had
returned the storage processor to its original position, we ran the ATF restore command,
which restored the system back to its original setting of using both storage processors.
RAID Level
Finally, we want to mention that although TeleStor allows for several different
RAID configurations, RAID 5 is by far the most recommended level for CTI applications.
(See the article entitled RAID In CTI, by Mike Bauer of IBus, Inc., on page 124 in the
May/June 1997 issue of CTI.) Used in conjunction with a Hot Spare disk (which
automatically kicks into action when a drive in a RAID group fails), it would provide for
a fail-proof storage system for missioncritical CTI applications.
CONCLUSION
If you are thinking RAID, we believe that TeleStor by Clariion is worth a look. TeleStor
is easily configured and maintained and it compatible with many popular operating systems
on the market. We also liked the ArrayGUIde utility, which allows you to monitor and
configure TeleStor from just about anywhere on the network. We think that ArrayGUIde
Managers user interface should more closely follow Windows standards and have an
improved online help facility, although we did find the program to be very intuitive.
On the hardware side, TeleStor sports an elegant but rugged look and is easy to work
with. There are plenty of redundancy configurations to choose from. The disks, storage
processors, and fans can be easily inserted and removed from their slots, and system
cabling to the server is relatively simple. We were happy to hear that work is under way
for an AC power supply integrated chassis.
The systems battery backup unit gives TeleStor enough time to shutdown in a
normal state in the event of a power outage. We can surely vouch for TeleStors
reliability as, during the week of our review, we were hit with numerous power outages and
brownouts and the system made it through just fine. As a final note, TeleStor does
generate a good amount of heat, so be prepared to turn up the airconditioning system a
notch or two. |