OpenPA.net

Stratus Continuum

Overview

Stratus Technologies built a line of Ultra High Availability Fault Tolerant PA-RISC (32-bit PA-7100 and 64-bit PA-8x00) based servers called Continuum (with a stated availability of 99.999% and more). These systems feature a great deal of redundancy, peaking in having four CPUs to form one single logical processor. The PA-RISC-powered systems were phased out in 2004 in favor of Intel IA32-based systems (not covered here).

The Stratus Continuum servers ran either Stratus-modified HP-UX 11.00, Stratus VOS (PA-RISC support from release 13.0 until 14.7.2) or FTX, Stratus’ own Unix System V Release 4 multiprocessor OS (however support differed between the 400 and 600/1200 lines).

Hardware Details

I/O

The Continuum 600 and 1200 series are basically the same system, except for the chassis configuration. The 600 has six slots for the main bus (called the Golf bus), and the rest of the space is filled with I/O card cages meant for secondary I/O boards. The 1200 has twelve slots for the main bus which occupies the entire width of the chassis. Secondary I/O boards go into a separate chassis. Both models have space for two rows of cooling fans on the top, and two rows of disk drives on the bottom (and also either a QIC or DAT tape drive or CDROM drive). The redundant power supplies with built-in UPS resides at the very bottom.

The main Golf bus is the main interconnect between the big boards. It is also redundant and self-checking. The big boards consist of the following (all these are FRUs):

On the 600 chassis, the six slots consists of two for the pair or CPU/memory boards, and four more slots for two pairs of big boards. On the 1200 chassis, there are slots for two pairs of CPU/memory boards and four pairs of big boards.

In addition the 600/1200 main chassis also has a pair of Console Controller cards which provides the RS232 console terminal and RSN modem connectivity. This controller also has a command mode that allows the operator to type commands on the console to reset the system, power down, power up, etc. It runs on housekeeping power that is independent of the rest of the system. The Console controller also contains some environmental monitoring circuitry that checks the chassis internal temperature and will increase the cooling fan speed if necessary (the fans themselves are also hot-swappable FRUs).

The secondary I/O chassis can be used to plug in a wide array of I/O boards (all Stratus proprietary). These boards are also used on the XA/R line. FTX supported many of the communications boards (ISDN, serial, parallel, X.25, and all sorts of other comm boards). HP-UX did not support many of those, if any. VOS also supported disk and tape I/O through this.

Expansion chassis is available to house additional secondary I/O cards or disk shelves (for large disk farms, etc).

↑ up

Architecture

Each logical processor is physically two pairs of actual CPUs (that means four physical CPU chips per single logical one).

Each pair is located on a separate FRU. All processors run lock-stepped (that is, they do exactly the same thing at the same time). Comparator logic between each two physical CPU pair monitors for discrepancies. If any physical CPU glitches or does something different, the comparator logic will detect the error and take that pair of CPUs offline, while the system continues to run on the other pair. There is no failover time. On multi-processor boards, each FRU contains multiple pairs of the logical processor halves.

The memory is self-checking and ECC corrected. If an uncorrectable error occurs, the FRU in which the memory is located will also be taken offline.

The DMA engines for the big I/O boards is designed such that the main memory content is protected from an errant card from scribbling over addresses that it was not supposed to write to. This of course is programmed by the OS device drivers.

The big I/O boards are also self-checking and contain a pair of everything. However, with the exception of the K600 they do not run lock-stepped to the twin FRU. For example on the K450/K460 boards, each of the SCSI host adapters is connected via the backplane into the same SCSI bus on the partner board, but each board’s controller occupies a different SCSI target ID. Only one controller is normally active, but when a failure occurs on the active board, all I/O is switched to the other controller. For the Ethernet ports on that board, they can be wired up to the same network or to different networks, and a software RNI (redundant network interface) layer provides transparent switching.

Other communications interfaces employ software-driven failover schemes.

All disks are mirrored. Early FTX 3.x releases used an in-house virtual disk layer (VDL) driver, but later releases switched to a modified version of the Veritas VxVM product. In HP-UX, HP’s own LVM (logical volume manager) is used. VOS, of course, has its own disk mirroring scheme.

The Continuum 400 series has the same CPU/memory architecture as the 600/1200, but the I/O bus is different. Instead of a Golf bus, it has an X bus that connects each CPU/memory module to a pair of PCI bridge boards. All I/O connectivity is via PCI cards. There are two PCI bays of 7 slots each, connected downstream from the PCI bridge boards. Each bay has a dual channel SCSI adapter on it as standard equipment. These are also cross-wired and dual-initiated much in the same way as the SCSI ports on the 600/1200 systems. The 400 is also typically shipped with a pair of Ethernet adapter cards. The PCI bridge boards also each contains a removable PCMCIA flash memory card. This is used as the boot device. FTX puts the bootloader as well as the UNIX kernel on there, whereas HP-UX only uses it for the bootloader.

The PCI bay doors control the power the the PCI slots. Once opened, all slots in that bay are powered off to facilitate removal and insertion of cards. The system continues to run on cards in the other bay. An interlock mechanism prevents both bay doors from being opened at the same time.

Again, all disks are mirrored as they are on the 600/1200 series, and communications interfaces use software-controlled failover mechanisms.

On the Continuum 400 the Console Controller is integrated into the CPU/memory FRUs. You can still reset the system via the software front panel on the console, but to power up/down you need to use the actual power switches on the machine (there are two, one for each power supply).

The Continuum 400 has two chassis versions, one is a short form-factor and AC powered only, and there is a tall CO (central office) version with a choice of AC or DC power.

Thanks to Ti Kan for the input.

↑ up

Operating systems

Operating system support was split between the Continuum 400 on the one hand and the Continuum 600 and 1200 on the other hand.

The Continuum 400 supported a Stratus-modified HP-UX Unix (11.00) and were marketed/offered with this system as main choice. The 400s also supported Stratus’ own FTX Unix, which was only sold on an exceptional basis however. There also was a cancelled effort to port the Stratus VOS operating system to the 400s. Continuum 400 servers running the Stratus-modified HP-UX 11.00 were fully binary (ABI) compatible with stock HP HP-UX, that is, programs compiled for normal HP-UX ran without changes on Continuum 400s.

The Continuum 600 and 1200 in contrast were primarily being sold with Stratus’ VOS operating system (non-Unix). VOS runs from release 13.0 until 14.7.2 on PA-RISC hardware.

“The original operating system developed for Stratus hardware was VOS. [...] VOS was designed from its inception as a high-security transaction- processing environment tailored to Fault-Tolerant hardware. It incorporates much of the design experience that came out of the famous MIT/Bell-Laboratories/General-Electric (later Honeywell) MULTICS project.”7

Also offered on the 600s and 1200s — on an exceptional basis — was Stratus FTX, the System V Unix from Stratus. Hardware supports was however more limited than that of the native VOS on these systems.

↑ up

System Table

Stratus Continuum PA-RISC servers overview
Model CPU Logical/
Physical
CPUs
Cache
per CPU
RAM
(max)
Expansion
(max)
Storage
(max)
I/O
(max)
OS
419 PA-8500
360 MHz
L1/P4 1.5 MB 8 GB 12 PCI 14 drives,
4 CD-ROMs,
4 tape drives
16 10/100 Mbit,
8 T1/E1,
64 Async,
64 RS232,
32 X.21,
32 V.35
HP-UX, FTX
429 PA-8500
360 MHz
L2/P8 1.5 MB 8 GB 12 PCI 14 drives,
4 CD-ROMs,
4 tape drives
16 10/100 Mbit,
8 T1/E1,
64 Async,
64 RS232,
32 X.21,
32 V.35
HP-UX, FTX
439 PA-8600
480 MHz
L1/P4 1.5 MB 8 GB 12 PCI 14 drives,
4 CD-ROMs,
4 tape drives
16 10/100 Mbit,
32 T1/E1,
64 Async,
64 RS232,
32 X.21,
32 V.35
HP-UX, FTX
449 PA-8600
480 MHz
L2/P8 1.5 MB 8 GB 12 PCI 14 drives,
4 CD-ROMs,
4 tape drives
16 10/100 Mbit,
32 T1/E1,
64 Async,
64 RS232,
32 X.21,
32 V.35
HP-UX, FTX
610S PA-7100
72 MHz
L1/P4 512 KB 128 MB 6 slots VOS, FTX
610 PA-7100
72 MHz
L1/P4 512 KB 512 MB 6 slots VOS, FTX
615S PA-7100
96 MHz
L1/P4 2 MB 128 MB 6 slots VOS, FTX
615 PA-7100
96 MHz
L1/P4 2 MB 1 GB 6 slots VOS, FTX
616S PA-8500
360 MHz
L1/P4 1.5 MB 0.5 GB 6 slots VOS, FTX
616 PA-8500
360 MHz
L1/P4 1.5 MB 2 GB 6 PCI,
2 Stratus I/O,
28 I/O
L47/P94 disks,
4 tape drives
10 10/100 Mbit,
8 T1/E1,
8 TR,
4 FDDI,
448 Async,
112 RS232,
28 X.21,
56 V.35
VOS, FTX
618 PA-8000
180 MHz
L1/P4 2 MB 3 GB 6 slots VOS, FTX
619 PA-8500
380 MHz
L1/P4 1.5 MB 4 GB 6 slots VOS, FTX
620 PA-7100
72 MHz
L2/P8 512 KB 512 MB 6 slots VOS, FTX
625 PA-7100
96 MHz
L2/P8 2 MB 2 GB 6 slots VOS, FTX
628 PA-8000
180 MHz
L2/P8 2 MB 3 GB 6 slots VOS, FTX
629 PA-8500
380 MHz
L2/P8 1.5 MB 4 GB 6 slots VOS, FTX
651-2 PA-8600
480 MHz
L1/P4 1.5 MB 4 GB 6 PCI,
2 Stratus I/O,
28 I/O
L47/P94 disks,
4 tape drives
10 10/100 Mbit,
8 T1/E1,
8 TR,
4 FDDI,
448 Async,
112 RS232,
28 X.21,
56 V.35
VOS, FTX
652-2 PA-8600
480 MHz
L2/P8 1.5 MB 4 GB 6 PCI,
2 Stratus I/O,
28 I/O
L47/P94 disks,
4 tape drives
10 10/100 Mbit,
8 T1/E1,
8 TR,
4 FDDI,
448 Async,
112 RS232,
28 X.21,
56 V.35
VOS, FTX
1210 PA-7100
72 MHz
L1/P4 512 KB ? 12 slots VOS, FTX
1215 PA-7100
96 MHz
L1/P4 2 MB ? 12 slots VOS, FTX
1218 PA-8000
180 MHz
L1/P4 2 MB 3 GB 12 slots VOS, FTX
1219 PA-8500
380 MHz
L1/P4 1.5 MB 4 GB 12 slots VOS, FTX
1220 PA-7100
72 MHz
L2/P8 512 KB 512 MB 12 slots VOS, FTX
1225 PA-7100
96 MHz
L2/P8 2 MB 2 GB 12 slots VOS, FTX
1228 PA-8000
180 MHz
L2/P8 2 MB 3 GB 12 slots VOS, FTX
1229 PA-8500
380 MHz
L2/P8 1.5 MB 4 GB 12 slots VOS, FTX
1245 PA-7100
96 MHz
L4/P16 2 MB 2 GB 12 slots VOS, FTX
1251-2 PA-8600
480 MHz
L1/P4 1.5 MB 4 GB 18 PCI,
6 Stratus I/O,
84 I/O
L95/P190 disks,
4 tape drives
18 10/100 Mbit,
8 T1/E1,
24 TR,
8 FDDI,
448 Async,
112 RS232,
84 X.21,
168 V.35
VOS, FTX
1252-2 PA-8600
480 MHz
L2/P8 1.5 MB 4 GB 18 PCI,
6 Stratus I/O,
84 I/O
L95/P190 disks,
4 tape drives
18 10/100 Mbit,
8 T1/E1,
24 TR,
8 FDDI,
448 Async,
112 RS232,
84 X.21,
168 V.35
VOS, FTX
  • Logical/Physical CPUs: L and P denote Logical and Physical devices — logical CPUs are made up of two pairs of CPUs each (i.e., 2×2)
  • Storage: L and P denote Logical and Physical devices — logical disk drives are formed from physical devices via RAID sets
  • I/O: Maximum number of I/O devices supported (not necessarily always configured with this number); notably the devices are also redundant

↑ up

References

  1. The Stratus Continuum Family (URL gone)
  2. The Stratus Continuum 400 Series (URL gone)
  3. The Stratus Continuum 600 and 1200 Series (URL gone)
  4. Stratus Machine History (URL gone)
  5. Stratus Continuum Series - VOS, Stratus Virtual Operating system (URL gone)
  6. Continuum 600/1200 Series (PA-7100) Service Announcement (URL gone)

↑ up