Conference 7.286::digital

    Patents are not generally awarded for features but for implementation
    of features. I would assume that it is specific algorithms that are
    patented not the features themselves.

    		Alfred

    For something to be patentable it has to be:

    i) Useful

    ii) Not obvious to someone who knows the field

    iii) Demonstrable

    One of the problems with the way patents (especially software patents)
    are currently issued is that someone who isn't familiar with the field
    (ie the US patents office) has to make a judgement call on (ii) above.

    You _are_ allowed to build on prior work, in which case the patent
    should be on the "improvements" over the previous design, not the whole
    thing. Part of the application process should be to identify such
    "prior art" (ie previously published material on related issues).

    John

    In my experience, the patent examiner does a key word search on
    existing patents, and if something looks similar to what you've
    done, rejects your patent.  It's up to you to prove that the
    patent examiner doesn't have a clue.  If you can't do that (or
    if there really is relevant prior art), then you have to reword
    your patent claims to limit yourself to something that is 
    distinguishable from the earlier patents.
    
    Anyway, this leads to a couple of results.  First, it leads to 
    incredibly impressive patent disclosures that are linked with
    extremely narrow claims.  Second, it leads to patents that are
    (or at least seem) carefully designed to not look like anything
    that went before, even though the ideas really aren't especially
    new or even very well thought out.  I've read a couple of patent
    disclosures that look like they were developed from cocktail napkins
    somebody picked up after some engineers had a party.
    
    Anyway, the fact that we have three patents on volume shadowing, or
    even the text of the disclosures, tells nothing about what we
    really got patent protection on.  Only the claims tell us that,.
    Did I mention that claims are written in a special language of
    their own?  Hard to understand, even if you wrote the disclosure!
    
    		Enjoy,
    		Larry

	Worldwide News                     LIVE WIRE

       Digital awarded three patents for Volume Shadowing technology 
 
         Digital today announced that it has been awarded three U.S. 
   patents for technology that provides customers with very high levels 
   of data availability in open client/server and time-share environments. 
   The patented technology is key to providing the level of data availability 
   and recovery performance needed by customers in business-critical 
   environments.
         The three patents (U.S. Patent Nos. 5,210,865, 5,239,637 and 
   5,247,618) cover specific algorithms and techniques that ensure data 
   availability and data consistency while maximizing I/O performance 
   in the event of system disruption or the replacement of a disk. The 
   patents are for Volume Shadowing, which offers Digital customers a 
   proven implementation of RAID technology for the OpenVMS Alpha AXP 
   and VAX platforms. 
         Under the patented technology, when a disk is replaced or added 
   to a shadow set, application access to data and copy operations 
   occur concurrently but with no extra synchronization. Copy operations 
   can transparently utilize disk controllers to make a new shadow set 
   member identical to other members, as quickly and efficiently as possible.  
   Use of host system resources is also minimized.  
         A merge algorithm is also covered under the patents.  If data 
   is being written to disk at the time a system disruption occurs, the 
   merge operations algorithm ensures that all shadow set members are 
   consistent. During merge operations, data remains accessible with 
   maximum application I/O performance. 
         The technology under U.S. patent #5,210,865 and #5,247,618 has 
   been issued in Australia; that under U.S. patent #5,239,637 has been 
   issued in Australia and Taiwan. Patents are pending in other countries.
         "The patents highlight the world-class proven functionality 
   that the OpenVMS operating system brings to open client/server 
   computing," said Don Harbert, vice president of Operating Systems 
   Engineering.  "As customers implement business-critical open client/
   server applications they need the combination of functionality, price/
   performance leadership, and support for multivendor computing and open 
   standards that the OpenVMS operating system provides."
         Volume Shadowing is currently available for OpenVMS VAX and 
   OpenVMS AXP systems.

Someone pointed me at this discussion. I was the project leader/architect of
Volume Shadowing and a co-inventor on all 3 claims.
(As well as other claims pending in the area of raid). For those really 
interested in the
gory details, these patents are for techniques 
described in the
Digital Technical Journal article I did on Shadowing's design in the summer 91
issue. 

The first 2 patents are not on the concept of  Merge and full copy operations.
They are for the unique algorithms we designed that allow
simultaneous updates by shadowing software (merge and copy threads) and 
application access thru all normal VMS mechanisms to the virtual unit with no
extra synchronization and which guarantees the correctness of both. This is a
hard problem, especially when clustering is factored in, and the solutions
meets the novel and unique criteria.

The third patent is for the mini-merge technology. This is combined shadowing
and storage controller technology that "piggy-backs" log and log control
information on WRITE commands. The controllers utilize internal memory to
store these logs under host control. This combination allows merge operations
to take seconds instead of hours with no extra steady state I/O operations. It
is state of the art in the industry, especially when coupled with a distributed
host based shadowing implementation. Other implementations in the industry
either 1) ignore the merge problem, 2) perform full merges across the whole disk
or 3) incur extra writes to ondisk data structures during steady state.

There is a fourth patent pending on the combined shadowing/controller technology
for direct disk to disk transfers for shadow copies, also with no extra
sybnchronization and correctness semantics.

Lastly, patents are written with many claims. The initial claims are very broad
and then narrow as you go on in the application. This is so that if a broad 
claim is
invalidated in court, the narrower claims can still be valid. So, for example,
the disk to disk copy patent starts out broadly claiming device to device
techniques and later claims narrow it to disks, and still later claims narrow it
to shadowing.

Scott

    Re .-1
    
    Congratulations on being awarded these patents. I think it is a shame
    that the LIVEWIRE people didn't include your name in the article. I
    think it is important to give credit where credit is due.
    
    I guess your personal name gives broad hints as to where your next set
    of patents will be coming from. I hope NT clusters are as successful as
    VMS clusters were in the 80s.
    
    Dave

Re:<<< Note 2812.5 by LJSRV1::DAVIS "Scott H. Davis - Windows NT Clusters" >>>

>The first 2 patents are not on the concept of  Merge and full copy operations.
>They are for the unique algorithms we designed...
>						...and the solutions
>meets the novel and unique criteria.

Thanks for the clarification, Scott (I was the one who asked the question in 
the basenote).  Having only once been involved in a patent filing back in the 
late 70's I wasn't fully aware of what's required today.  And back when I wrote 
software, patenting any of it was not a possibility...  :-)  :-)

I'll look forward to reading "the gory details" and must add my congratulations
on your accomplishments.

-  David

Thanks for the congratulations. I and my other partners in crime were also 
disappointed that our names were not mentioned. It isn't everyday that patents
are awarded... :-(  Sigh...

The other inventors, in addition to myself, on all the Shadowing patents are:

	Bill Goleman 
	Dave Thiel

And on the 2 patents with storage controller functionality (one granted, one 
pending) add:

	Bob Bean
	Jim Zahrobsky