Passionate About Tandem Ross Systems International Limited
Tandem/HP NonStopTM
Development & Test Solutions |
|
Mistley Telegram
January 9th,
2012
Happy New Year to you all! As a contractor looking for work the first thing I did was to look at
various job sites amongst which was to
I have been following Real Time View including the link "Is NonStop alive and kicking as a viable platform..."
I note with considerable amusement that several of
the participants think that the NonStop OS is rather old fashioned since the
first versions came out in 1975, well:
No, I think that what the most people are thinking is
that NonStop does not have a GUI front end as standard, whereas at the same
time not realizing that all the GUI front ends, except maybe Mac, have been
built as front end lash on's to much more ancient OSs.
The fact is that Guardian is still very modern and
when it was built was so far ahead of the competition that they were left
trailing in the distance. This needs to be emphasized by NED and also a
platform wide GUI front end, which uses TCP/IP and SPI to communicate with the
underlying OS is not a bad idea. It also needs to be made backwards compatible
to S and maybe even K series, no big problem seeing the longevity of the SPI
interface, it is just a question of shutting out functionality as the product
regresses back through the OS series. N.B. This is what we are doing with our
latest version of FINFO, it has a Java front end and a multithreaded TCP/IP
listener linked to the FINFO library at the Tandem end.
I suppose you could achieve this by a
comprehensive expansion of the web viewpoint functionality, to provide a
universal interface, which would be much better than the ad hoc expansion of
GUI interfaces to NonStop which we have at present.
Sorry to rattle on about this one, but the
comments on Real Time View got my goat, so as to speak, but the authors were
not there when it happened, so how could they know!
;-) Rupert
February 17th, 2011 BITUG SIG At the start of December last year there was the BITUG BIG SIG in Trinity House. It was the time of the big freeze and so many people who intended to attend were unable to do so. However, this did not stop many of HP leading lights in the NonStop arena keeping their promises to attend, including Bob Kossler, who I had a chance to chat with about our products and the problems in demonstrating any NonStop products, since it is virtually impossible to set up a demonstration system on any temporary site and on demand communication to NonStop systems is fraught with difficulty. As a result of this it occurred to me that for demonstration purposes these difficulties could be got over if a simple interactive replay program could be written to give the impression of a NonStop session and I pencilled out a few ideas on the spot. For instance such a program would have to have two modes and input mode, where the replay file was displayed character by character on the screen, and a display mode in which the replay file was simply displayed on the screen without interaction. Also to give realism and to go over what had previously been output the program had to be sensitive to Page Up and Down when in the input mode. As a result of this I developed the Tandem Replay product for demonstration purposes and because time was short, the initial version runs under MS-DOS on a 24 bit screen, which means that it is useable up to Windows XP, and I gave it a 1000 line paging buffer, which appears to about right for any demonstration purpose. The Input file is captured from the session stripped of all special characters and STX (0x1) Inserted after prompts when input mode is required and ETX (0x3) Inserted after the input has completed, immediately before the response to be output. The program naturally terminates on the end of the input file or receiving an EOF (Ctrl C) from the keyboard. I used this at the next SIG to demonstrate the changes which had been made in FINFO on the 16th February 2011 to the Operations Management and Systems Management SIG at 88 Wood Street in London which was kindly hosted by HP and ably managed by Ian Liston-Brown from HP and Neil Barnes and Matt Whiteman from BITUG. It appeared to be well received but there was a problem with the overhead projector setup, when I tried to enter full screen DOS mode. Thus I had to use the command prompt windows mode which looked a little tacky, but never mind. However, if you would like to look at this demonstration again please download the replay and the capture file, start a command prompt screen and enter REPLAY BITUGFEB.TXT You can then step through the demonstration by pressing any key, using Page Up and Page Down keys for paging through the displayed text. You can exit at any time by pressing Ctrl C. The capture program will work until 28th February 2011 and you can also use it to replay your own captured files, annotated as described above. October 6th, 2010 FINFO 3.4 There are times when a product seems to come alive and take a direction of its own. Such is the case with FINFO. I remember only to well standing up before BITUG and saying that the natural course of things is data produces information which then produces action. Never, realising the events which would unfold this year. There have been several conversations this Autumn about using the information to be gained from FINFO to steer NonStop systems and it appeared useful to prototype these ideas before committing them using a SPI interface into the various system programs within the NonStop architecture. This task has been made much easier by the fact that every utility on the NonStop platform has both a command line and a SPI interface and the command line interface always has an obey file capability. Thus it is possibility to produce obey files to configure the system from FINFO by means of a fairly simple change and also then using this functionality to prototype the corresponding SPI functionality for a whole range of NonStop system applications, more about this later. The second hurdle which needed to be crossed was to adequately answer the question of whether the file selection parameters within FINFO were adequate for this task? Having taken advice the answer to this question was a conditional YES except I have not allowed for negative conditions i.e not safeguard, not file codes... Having taken this step, for completeness, I also gave the file sort command -O the capability of reverse sorting. The changes were implemented as follows:
All of this is a preamble to round trip engineering for SAFEGUARD and other products, since it enables a command/obey file to be generated from the current disposition of the data store infrastructure. The next step is of course to put the safeguard SPI command interface in and to develop further flags for building the ACLs etc.
Thus, it is now possible to express in a FINFO command
line the following:
FINFO -S<cond1> -!S<cond2> ... -CB1"ADD " -CA1"<user
permissions> &
-CB2"ALTER " -CA2<acl 1> -CB3"ALTER " -CA3<acl2> ...
which would result in the generation of for each file
found
ADD <file> <user permissions>
ALTER <file> <acl1>
ALTER <file> <acl2>
...
This can be fed straight into SAFECOM to set up the
safeguard protection of the system, it can also be envisioned that this
functionality could also be used to feed information into EDIT, FUP and other
HP NonStop utilities..
It is a small change in FINFO but it has the capability
of saving system administrators a huge amount of time in securing a system
containing many different types of files for which the FILE-PATTERN directive
is not adequate.
What we have here is a full patterns directive over the entire range of file attributes which gives the system administrators a much finer control over securing the system, without the hassle of having to repeat the setup for every single file. However, the fact that it is possible to generate control information for not only SAFEGUARD but also for a whole host of other HP NonStop utilities raises the possibility of linking FINFO using the SPI interface to all these utilities and of generating a very comprehensive dashboard facility based on it. In conclusion this is good but it will get better with the the SPI interface(s), round trip engineering and the GUI interface to go. Life is getting exciting! So, request a test licence for FINFO Version 3.4 (S-Series) or 5.1 (Itanium & CISC), available now. September 21st, 2010 SafeGuard Futures Our article in August's Tandem World was very well received with a huge number of readers, Thank You. We also receive some very valuable and interesting feedback the gist of which was that it was not just the determination of how the system was protected by SAFEGUARD, but also how to set up SAFEGUARD on a database. The general moan seemed to be even with the most recent point and click interfaces and the inclusion of DISKFILE-PATTERNS in the SAFECOM and SPI interfaces the job of protecting a system using SAFEGUARD was very messy and did not really represent the way in which most system managers think. The problem can be likened to trying to paint a picture and rather than having broad brushes to spread many different types of paint on over the broad range of the canvass one is limited to having a small brush which can paint only one pixel at a time and a paint a square tool. The kernel of the matter lies in the low level nature of the SAFEGUARD descriptive language whether expressed directly or in the form of point and click favoured by many GUI interfaces which are all constrained by the problems of granting or denying access to a file to a given user under a given set of conditions. The is also the matter that the SAFEGUARD and Guardian environments are seen as distinct whereas in reality at the file system level they are tightly integrated. What is really needed is a high level descriptive language where once could say for instance:
FOR ALL SQL FILES DISKS CALLED
name_pattern_list
Or Or something like that and then having an engine which could compile this into the corresponding SAFEGAURD commands. It would also be even more revolutionary to have a decompiler which could look at a system and generate a description of the SAFEGUARD status of the system from what it found: ALL DISKS NOT SAFEGUARD
PROTECTED The terms round trip engineering and object modelling spring immediately to mind with the quantum leap which always accompanies the transition from purely context free declarative and imperative languages into context dependant natural languages of which our own is an example. It also has the property of combining the GUARDIAN and SAFEGUARD environments at the top level of expression. We are working with our partners to further this idea and decide what is required for it to be a success. Your ideas too would be more than welcome. In the meanwhile you will simply have to manage with the existing facilities which are contained your favourite package or in our current version of FINFO as shown by last month’s white paper. August 18th, 2010 FINF0 Version 3.3 Enabled for SafeGuard File Reporting (now available) After considerable consultation and much work the FINFO Version 3.3 is now available. The chief change from earlier development versions are:
This has now been done (and tested!) and so the whitepaper has been updated with the changes, so if you want to see what has changed you will need to download it again. I have also updated the User Manual with the changes and it is now available for download. Postscript (19 Aug 2010): TANDEM WORLD managed to mangle our article: It should have looked like: Ross Systems International SAFEGUARD™ Reporting Ross Systems International, have added comprehensive SAFEGUARDTM reporting to their disk file reporting package and have produced a White Paper giving details of the new functionality. Title: Reporting HP NonStop (Tandem) SAFEGUARDTM Protected Disk Files Using FINFO Subject: HP NONSTOP; FILE SYSTEM; SAFEGUARD; SECURITY; Author: Rupert Stanley, Ross Systems International Date: 8th August 2010 Audience: SECURITY MANAGERS AND SYSTEM ADMINISTRATORS Status: VERY URGENT Abstract: SAFEGUARD is an HP NonStop (Tandem) proprietary product which is used to provide an extra layer of security to HP NonStop objects, such as Users, Processes, Devices and Disk Files.
1. Access Control Lists, which contain information as to whether permission to access the objects is to be granted or denied based on user (group, number) and operation (read, write…). 2. File Transaction Auditing Conditions, based on type (Access / Management), Outcome (Success or Failure) and Source (None, Local, Remote or All). 3. File Status Information. i.e. whether the file is Licensed, ProgId’d, to be Cleared on Purge of if it is Persistent This facility is very important is securing sensitive systems, particularly in the financial service industries, where there is a statutory obligation, and it provides a very competent facility for meeting these obligations. Although the Auditing facilities are excellent there are several operational challenges in the status reporting mechanism which extensions to the RSI file information product FINFO address. This white paper addresses these problems and shows how they are solved by FINFO Version 3.3. Download: WHITE PAPER Product: FINFO Version 3.3 Availability: Now. Trial: 30 day trial license available on request. Email: Rupert Stanley, rupert@rsi-ns.com Telephone: +44-1206-392923 Web: http://www.rsi-ns.com/.
August 9th, 2010 FINF0 Version 3.3 Enabled for SafeGuard File Reporting (the continuing SAGA) Overnight my helpful readers suggested that I should put the global SAFEGUARD settings into FINFO. This has now been done (and tested!) and so the whitepaper has been updated with the changes, so if you want to see what has changed you will need to upload it again. August 8th, 2010 FINF0 Version 3.3 Enabled for SafeGuard File Reporting As I said in my last report work has been progressing apace on the safeguard reporting functionality, all the design and programming work has been done an the new version is in the last throes of testing and documentation.. It is also a very important release because it addresses several of the problems which are encountered when using SAFECOM to access information for reporting on SAFEGUARD protected files. The new functionality addresses the requirements for a decent and intuitive on screen reporting package as well as the need to export SAFEGUARD data onto the PC platform for further processing by Excel or Access. The chief strength of the product lies in it being able to be aware of all the various layers of disk file protection afforded by the linked objects and ACLs and produce report data which expresses these links, which are so important in securing the HP NonStop Data Store. It is also very convenient to be able to select SAFEGUARD protected by that attribute alone. The features of this new version of FINFO are summarised in a downloadable whitepaper. I look forward to your comments and also it available on 30 day trial on application. June 25th, 2010 BITUG Presentation Follow Up / The Other Half It is strange how when you set off in one direction you are often waylaid by events and finfish up by going in completely the opposite direction. It was thus with the BITUG presentation of 25th May, whereas I have written up in quite considerable detail the FINFO component. It is however the HSM management element of the presentation which is receiving most interest. This is not really surprising however, since HSMs are expensive beasts, about £50,000 a pop, and together with their maintenance costs at between 20 and 25% per annum, i.e. £10,000 each, consume considerable sums of money. Add onto this the costs of the HSM drivers, which I have had intelligence of as costing about £100,000 per annum for an institution running a ATM and/or POS payments package. Thus a big financial institution will probably have about 7 HSMs for which the costs look like:
This means that the setup price is £350,000 plus, especial if the driver software needs modification, and The annual maintenance costs, excluding staff is of the order of £170,000 per annum and These costs are very much dependant on the number of HSMs used and do not include staffing or other costs. No wonder that there is great interest in anything which can bring this enormous cost of ownership down. However in order to do this and to provide the service which financial institutions are legally and commercially required to they need to know the processing capacity of their HSMs in real terms for their transaction mix and load, and the only way of doing this is to be able to monitor transaction speeds from HSMs in the running application. This is something which the existing HSM drivers are unable to do and which is already present in a simple form in our existing products and for which it is a fairly simple task to RPQ up to provide full reporting either in the form of extractable periodic reports on on-demand periodic load analysis. The possible solutions are included in the discussion document which should be read in the context of our white paper on HSM Farms. FINFO Finally, the work on incorporating Safeguard information into FINFO is progressing apace with the writing of the SAFEGUARD SPI interface and has reached the point where the output looks like: $WORK NATIVE 4> finfo -g FINFO V2.8 Native 24/06/2010 18:53 ----- Copyright Ross Systems International Ltd. 2008 In House Version(Release Date 5th January 2009, Never Expires
Files on \SIRIUS.$WORK.NATIVE Audit Safeguard Access Manage Name Last Modified Size Status LCPC PassFail PassFail Owner RWEP ZZVS00T 15-Oct-2009 11:13 57,344 Thawed YNYN ALL REM LOC ALL 127,001 **** Access Control List Owner 127,004 RWEP 127,005 R E 127,006 DENY E ZZVS00X 14-May-2010 12:13 14,336 None 127,001 NONO ZZVS00Z 24-May-2010 12:39 14,336 Thawed YNYN ALL REM LOC ALL 127,001 ****
Selected User Totals for SubVolume \SIRIUS.$WORK.NATIVE User No User Name Files Bytes Used Pages Used 127,001 RSI.RUPERT 54 4,436,464 2,578 Totals: 54 4,436,464 2,578 Watch this space, I still have the subvolume and volume analysis to do! May 25th, 2010 BITUG Presentation On 19th May I was invited to give a presentation of our Systems Management and performance tools. As a result of which I decided that presentations of our FINFO and HSM Farming solutions would be of interest to the attendees, for a half hour slot this was a challenge. BITUG as its name suggests is a Tandem User Group, which in reality means system managers, with a very thin sprinkling of business users and system developers. This means that any presentation must be focused on their perspective and requirements. A somewhat difficult task for a dyed in the wool developer like myself. So, I planned to approach the presentation with the general plan of: DATA → INFORMATION → ACTION and started to write the presentation on this basis with nice slides to ram the point home. However, as they say there's many a slip betwixt the cup and the lip and as it so happened I had a very urgent non-moveable appointment sprung on me for the 18th. Oh how history repeats itself. The end result of this was that the message got slightly mixed, since I was still in preparation at 5 a.m. on the morning of the presentation and had to hurry. Also maybe it would have been a good idea just to have dropped the HSEMM component, but there is always a temptation to get as much in as possible, which is not a good idea,. I now know that the message should have been: FINFO TAKES THE DISK FILE DATA
What could be simpler? I should also not have muddied the waters with information about HSM farms, which although is very useful information for system managers is a whole one hour session on its own. At the end the chief comment made was that FINFO is a good tool because:
The audience would also like to see SAFEGUARD information added and look forward to seeing the GUI client version to bring it up to standard data-centre presentation standards. Work which is now in progress. The comment was also made that the product is a dream ticket for developers because it would make life so much simpler for them and save them huge quantities of time and money with its advanced file search and display facilities, but I needed to get to that audience, and Finally I did not mention that the first of our clients using the product were saving considerable amounts of money in the systems management group compared to tying to do the tasks assigned using TACL macros, as they previously had done. Anyway, I hope if you attended you enjoyed it and if you did not then you can download the notes from our web site. download. Thank you for listening and for your useful advice. If you would like to try a full function demo version please contact me rupert@rsi-ns.com FINFO In the meanwhile FINFO has been in an extended Beta phase. I have given it more functionality in that it now gives a detailed set of recommendations for action after the sub-volume summary display. I have also fixed a couple of bugs which our user base have found, as usual it is always in the most abstruse parts of the functionality that the bugs are uncovered. I suppose that the more a section of the code is exposed to the user base the more likely it is that problems will become apparent when it is used. There was also the fact that the multi-threaded server needed to be built and tested, which took some time. I am also still working on the Java client server and will try to also include the comments made at BITUG concerning the incorporation of SAFEGUARD functionality. So, all is going ahead but somewhat slowly. April 5th, 2010 Site Errors I have just been informed of a number of missing links in our site. The problem with the index page has now been resolved and the probe link below now works. Sorry folks. March 29th, 2010 FINFO 3.2 At the current moment I am in the process of developing a SPI version of FINFO with a corresponding TCP/IP interface and JAVA front end and am now testing the Tandem components whilst at the same time building the Java components. As a result of this I have been made aware of the fact that the Java components need to navigate through the Tandem volume/subvolume/file hierarchy, a facility which is at present not available in the stand alone versions of FINFO. So there was a need to go back to the drawing board to include this capability and since I want to make the product consistent, it was therefore necessary to include this in the base line versions. I considered the best way of including this within the structure of the FINFO command set and I came up with the conclusion that it would be best to be able to type in: FINFO $<vol> to get the volume information, wherr <vol> could be wildcarded, and FINFO $<vol>.<sub> to get the subvolume information where both <vol> and <sub> can be wildcarded The question was then how was the information to be displayed and I came to the conclusion that for the volumes display, it would be a good idea to follow the DSAP ,SHORT format as follows: >FINFO $* \sirius.$work.native.licenses Note the inclusion of system totals However, subvolume display offered a possibility to perform an unique analysis of disk usage which was far more valuable than the straight listing afforded by FUP including an analysis of what percentage of the disk space was being used by the various subvolumes and if any files were at their limit, identifying where they were situated:
>FINFO $WORK.*
\sirius.$work.native.licenses PCPROGS 101 0
1 4.84 0.13 2796 ZSPIDEF 510 0
3 22.96 0.88 18402 Note. The subvolumes where there are files at their limit are identified and the amount of the disk estate consumed by each subvolume is identified together with the number of users of the disk. The system total for the subvolumes analysed is also included, including an average usage across the entire disk estate. Thus converting a simple information tool into a hot tool for system space usage analysis, something which to my knowledge is not available elsewhere. The development of FINFO will continue with the Java front end and SPI server. However, in the future the price of the product, which will include this functionality, will reflect the added capabilities. However, if you become a registered used before the Java front end functionality is released you will receive upgrades at no additional cost, including the Java front end release, and the future pricing of the product licenses for your systems will only reflect the small annual increase for the non Java version, in spite of your receiving all future upgrades including the Java releases. To order go to the FINFO Order Page |
Contact:
info@rsi-ns.com Tel:
+44(0)1206-392923
Copyright © 2006-2010 Ross Systems International Ltd.
Registered in England No.2407494
|