Tony Harminc wrote:

> Sigh... I really don't want to turn this nice quiet list (that I've been a
member
> of for many years) into a bunch of vendor arguments. It's rude, and in this
> case it's almost 100% off-topic.

OT? (With respect, Sir, the subject of this thread is "RSA SecurID
product.")

My last post was written as a direct response to Mr. Harminc's comments
about RSA's SecurID,  a one-time password (OTP) token that is sold in
direct competition to the Vasco OTP tokens that Mr. H's employer
bundles with his SSO products.

Citing facts not otherwise available to the List is generally
considered to be a constructive contribution to a discussion about
competitive products.

Mr. Harminc's initial post was, IMNSHO, fundamentally misleading
because it fostered the false impression that all hardware OTP tokens
are the same -- that all offer comparable degrees of security,
integrity, and process assurance -- just because they all manage to
cough up an OTP... one way or another.

I offered additional information about RSA's SecurID -- and about the
Digipass token Mr. Harminc's employer resells for Vasco -- because that
seemed to be the proper way to counter Mr. H's little list of Issues of
Concern for potential OTP buyers, which I believe was biased against
RSA's product.

The fact that, according to the CEO of Vasco, many European financial
institutions have recently decided that they will have to replace their
Vasco DigiPass tokens every 12 months  -- when millions of those tokens
were originally purchased and distributed with the expectation that
they could be used for many (3-7?) years -- seems to me relevant to any
discussion of the various 2FA/OTP options.

(I do apologize for my error in describing Mr. Harminc as a developer
of software applications for Vasco. Mea maxima culpa, I inadvertently
cut a line. Mr. Harminc is a developer of SSO applications, which are
sold bundled with Vasco Digipass OTP tokens, for Proginet
<www.proginet.com>. Proginet apps, said Mr. H, allow Vasco tokens, or
any Radius-compatible 2FA solution, "to be authenticated against
mainframe security systems (RACF, ACF/2, and Top Secret.")

Mr. Harminc reviewed the comments of Vasco CEO T. Kendall Hunt last
summer at <http://tinyurl.com/cospv> and wryly flicked his FUD-O-Meter:

> I had never heard of Mr. Hunt, but I went and read the article. In it, Mr.
Hunt
> says that "most banking organizations using Digipass tokens refresh the
>  products about every 12 months for improved security". This seems
> unremarkable, despite Mr. McLellan's attempt to make it sound like some
> dark secret. Organizations that deal with Real Money will have a threat
> model quite different from those using tokens to protect other kinds of
> assets, particularly when the user base in question is essentially untrained
> in security awareness.

Unremarkable, huh?  I'd love to hear the ruminations of some savvy
Proginet customer, who bought your bundled Vasco token package, as he
mulls the implications of the fact that the top European banks have
decided, for reasons unknown, that they must now spend maybe 5-7 times
what they originally budgeted for tokens to cover the unexpected costs,
over the tokens' projected lifespan, of annually purchasing and
redistributing new Digipass tokens to their staff and customers. ;-)

Is there anyone who thinks these banks are spending money for the sheer
joy of expenditure?  Is Ebenezer Scrooge a miser? Is there anyone who
can be utterly certain that their local environment is exempt from
whatever new threat the Euro banks are reacting to?  (And aren't *all*
users "essentially untrained in security awareness?")

"Dark secret?" <Sigh.> Gimmme a break!

What Mr. Hunt publicly reported was an unexpected, unexplained -- and
very expensive -- change in the security policies and the provisioning
practices of many heavily-regulated European banks.  Yes. it plays hell
with the Total Cost of Ownership for Vasco and Proginet products -- but
there may be bigger issues here too.

Institutions that "deal with Real Money," and online banks in
particular, do indeed have a different threat model than the rest of us
-- but those differences may not be what's at issue here.  If it is the
OTP tokens, per se, that are the focus of the banks' operational risk
concerns, whatever new information those banks have pugged into their
Basell II models may be of interest to us all -- not just Vasco
customers.

To judge by anecdotal evidence, tokens -- some labelled valid until
2008 -- are being exchanged one for one, with identical replacements,
by some big European banks. Security officials at these banks are being
unusually reticent in explaining their new policies, which has led some
European observers to speculate that they are operating under
unannounced regulatory mandates or EC guidance. (If that is FUD, I
apologize. I expect more details will be made public, sooner rather
than later, but that's all I know.)

I don't want to be Cassandra here. There are a lot of exciting things
happening as the big 2FA vendors, RSA and Vasco in the front ranks, get
pulled into the global drive to supply a billion dollar mass market for
2FA in retail financial services.  (At least for RSA, btw, that may be
a largely mainframe-focused marketing initiative.) New vendors, some
with exciting new technologies, are popping up all over.

RSA's recent purchase of Cyota, a developer of layered anti-fraud
technologies; OATH's RFC on mutual authentication, and Vasco's new
strategic tie with Tri-Cipher, are all artifacts of a broad competitive
realignment of AAA vendors and 2FA products, which is likely to
redefine identity & access controls, provide much more secure
transaction on the Web.  Six months from now, the service offerings in
this industry niche are going to be amazing. ;-)

Vasco's report of a new caution, a new skepticism, about their OTP
tokens among top European financial institutions is a bit of the fly in
the ointment -- and while I'll freely use Mr. Hunt's comments to ward
off Vasco enthusiasts, I also admit that I am very curious about what
is happening there.

Earlier, I suggested -- admittedly on scant evidence -- that the most
likely stimulus for a drastic European reevaluation of the risks
associated with some OTP tokens is probably the European Commission's
major research project on Side Channel Analysis (SCA) and various
SCA-based attacks on embedded cryptosystems.

Two years ago, the EC funded nine separate research teams, based across
Europe, to explore the threat of SCA attacks on microchip-based crypto
implementations -- and to seek defenses, in microchip design, against
specific attacks, like Differential Power Analysis (DPA). The SCA R&D
project <http://tinyurl.com/7zfaa> is due to produce a final report at
year end, but it has spun off numerous research papers, and more than a
few rumors, over the past two years.

This is not OT. This is relevant to this discussion -- whether I'm
right or wrong to make a link between the SCA  threat and the European
banks' 12 month turnover policy on tokens -- because the SCA threat is
real, and because RSA's AES-based SecurIDs, introduced four years ago,
were specifically designed to be resistant to SCA attacks.  Most other
OTP hardware tokens, including Vasco's, are apparently still vulnerable
to this type of direct attack on the token's OTP processing circuit.

Guarantees that a personal authentication token is "tamper-resistant"
are often sought by CIOs, ISSOs, and network managers who are
considering 2FA options.  It is, and should be, a relevant issue in any
discussion of 2FA options and hand-held personal authenticators like
RSA's SecurID.

I readily concede, however, that this is a high-assurance issue.
Tamper-resistance is too often assumed for all hardware OTP tokens,
when it should not be  -- but it is also perhaps not a critical issue
for many institutions considering 2FA.

For many installations, cell phones or PDAs, running token-emulation
software, provide a level of secure 2FA wholly appropriate to their
threat and risk environment.  (At worst, hardware OTP token vulnerable
to DPA attacks would fall roughly in this category -- with the 2nd
factor, the user-memorized PIN or password, still providing a
protective barrier to illicit access even if the "token" is stolen or
compromised.)

Other institutions -- probably including the federal agency whose
interest in RSA's SecurID sparked Mr. Harminc's avuncular guidance --
arguably have more demanding security requirements.

It makes no sense to spend more on security than the value of the
resources at risk, but there are obviously institutions -- including
many which do not deal in either Real Money or Dark Secrets -- which
have data and other networked resources of such value that they
routinely demand, and require, a high degree of assurance for all
elements in their security infrastructure.

Not all 2FA solutions are equally secure; just as not all tokens are
created equal -- but most 2FA vendors provide an array of software and
hardware options to meet these different requirements.  Sites with
different needs require different levels of integrity assurance for
their critical infrastructure.

Given the ease with which malware continues to spread and corrupt local
workstations, there is likely to be renewed interest in the integrity
of authorization models, intrusion detection devices, and the forensic
audit trails.  In this, the mainframe world is far ahead of the general
market -- but even the regulars on this List are likely to see
intriguing new techniques for adaptive, upgrade-on-the-fly,
authentication services emerge from financial services, where the
losses (cash and confidence) now associated with phishing and identity
theft are simply unacceptable.

Ok, back to our gentlemanly dispute.

Despite the fact that Mr. Harminc concluded his original post with an
apology for a "sales pitch," Mr. H continues to insist that my
suggestion that his advice for OTP newbies was biased is "simply
absurd."  For the entertainment of the Listocracy, here's Mr. Harminc's
original worry list:

>> Do they make their money on the actual tokens, or on proprietary server
>> software, or on some other basis? Can you buy additional tokens from
>> another source or are you locked in? Conversely, can you use their tokens
>>  with another server? Is the server priced by user in addition to the cost
of
>> the tokens, or by number of connected endpoints, or some other model?
>> How long do tokens last, i.e. do they expire by design before their natural
>> battery life? etc. ect.

In his latest post, Mr. H added:

> May I gently suggest that re-reading my actual post, rather than some
> phantom one, is in order... I said nothing at all about SecuriD's "unique
> capabilities", let alone cast them in negative terms. I have no idea what
> magic patented functionality only RSA can provide...

What, pray tell, other than the SecurID's "magic patented
functionality," could be the basis for the fearful "lock in" Mr.
Harminc cautioned prospective OTP buyers to avoid?  ;-) Given RSA's
relative success, if the SecurID could have been cloned by its
competitors, it would have happened long ago.

The  List's readers can  judge for themselves the validity and
"objectivity" of Mr. Harminc's concerns about the risks associated with
an institution's decision to commit itself to a portfolio of patented
2FA technologies that, admittedly, can only be purchased from one
vendor (RSA, the patent holder); and which, admittedly, can only be
supported by the authentication servers sold by one vendor (RSA, the
patent holder.)

Proginet, Mr. H's employer, and Vasco, both offer products that rely
upon the basic Radius protocol and a Radius AAA server to manage their
authentication calls. RSA's SecurID, admittedly, offers a proprietary
udp protocol -- which encrypts the authentication exchange between an
app-integrated RSA Agent and the RSA server -- to support a proprietary
authentication engine, the RSA Authentication Manager, on a variety of
hosts. Of course, the RSA AM also includes a Funk commercial Radius
server.

Mr. Harminc contrasts the two architectures in economic terms,
highlighting the fact that several Radius servers are open source.
It's a worthy point. I thought it useful and appropriate, however, to
also point out that over 300 networked devices and applications, most
with integrated RSA Agents, ship "SecurID-Ready" -- with certified
out-of-the-box compatibility with SecurID 2FA.  Those partnerships have
been a major factor in RSA's 15-year domination of this market.

In the same economic context,  Mr. H's expressed concern about the
respective business models used by RSA and other 2FA vendors, Vasco
among them: the balance of revenue they draw from tokens, agents, and
servers; and whether the OTP infrastructure is priced  by user (as
RSA's is) or on some other basis.

All vendors need revenue and profit; but revenue is like digital data.
it can be channeled and poured from many different vessels. Mr. H and I
could debate the particulars of various business models -- but those
seem to me details more properly left to competitive salesmen
discussing a specific network.  For any given purchase decision, it's
obviously necessary to compare prices, but any discussion of how
various pricing models effect the cost of a 2FA implementation really
should be site-specific.

And ultimately, the acceptability of any given pricing scheme is
decided in the competitive marketplace, isn't it?

Mr. Harminc, above, cautioned prospective 2FA buyers to study:

>> How long do tokens last, i.e. do they expire by design before their
>> natural battery life? etc. etc.

Much of my original reply to Mr. H was simply an exploration of that
"etc. etc." ;-)

It is true, of course, that RSA's SecurIDs -- which are always on;
continuously displaying their 60-second OTP token-codes -- have a
sealed capsule, and are sold with a pre-programmed lifespan, to a date
certain, of 1, 2, 3, 4, or 5 years. (The SCA threat, as it gets more
widely acknowledged, may underscore the value of a sealed token.)

It is also true that most competitive OTP vendors, like Vasco, sell
"event-synch" (click for an OTP) tokens with either (a) a replaceable
battery for eternal life, or (b) a sealed capsule which, with a click,
will produce OTPs on demand until the token's internal battery gives
out, nominally somewhere between three and seven years.

These have been basic Facts of Competition in the 2FA token market for
well over a decade -- but I have yet to understand the implied moral
superiority in Mr. Harminc's suggestion, echoed in bales of competitive
literature, that RSA is somehow cheating, or doing a disservice to its
customers, when it sells SecurIDs that "expire by design before the end
of their natural battery life."

The SecurID tokens are only one element in RSA's authentication
service, which is packaged and priced, per year, to meet the needs and
expectations of its customers. Historically, cost is only one of many
factors involved in a decision to purchase critical infrastructure --
and seldom is it a determining factor.  RSA is rarely if ever the
lowest-priced option, but RSA's choice of design, functionality,
pricing, and packaging -- buttressed by its partnerships, and its
extensions on its basic 2FA services for everything from SSO, to
Federation, to Windows XP logons -- for its SecurID infrastructure has
generally found favor in the marketplace.

With over 19,000 SecurID installations world-wide, as I noted, RSA is
estimated to supply some 70 percent of the enterprise demand for OTP
tokens

But what's the OTP product and what's the authentication service? And
is the distinction critical, or even useful, as the global market
evolves toward a networked service model?  While a number of other OTP
vendors are trying to standardize around OATH's SHA-1 design for a
single token, RSA is arguing for cipher agility, fostering discussion
of high-assurance certs among CAs; and publishing a series of
standardized templates which offer developers guidance for securely
integrating OTPs and 2FA validation into a wide variety of emerging
products. See the One-Time Passwords Specifications at:
<http://tinyurl.com/3w7c7>

To meet the needs of financial institutions worried about consumers
being forced to carry multiple SecurIDs, RSA and eTrade Financial, the
online brokerage, are now beta testing a new Internet-based
authentication service  through which RSA will only validate a SecurID
"token-code" for financial institutions which are willing to accept and
co-validate, with a separate password, 2FA from a consumer holding a
SecurID originally issued by another institution.

(RSA will validate the OTP, for a user previously registered, by
SecurID serial number and a PRN pseudonym, by an authenticating
institution -- but only that financial service firm or bank will have
access to the account information associated with the authentication
call.  In this limited service role, RSA's can't even identify the
individual; it simply blindly validates the OTP factor for the
inquiring financial service firm.)

Given the range of RSA's initiatives in products and services, it's
hard to go "off topic" -- but it's been a quiet afternoon, and I
probably pushed the envelope. I beg the indulgence of the Listocracy
for the length of these comments, and some admittedly OT ruminations.
Withall, I submit the facts that I reported about the relative
integrity of the various OTP tokens -- in particular, their respective
resistance to direct but perhaps non-destructive physical attacks,
using side-channel (SCA) attacks -- are at least as relevant as Mr. H's
niggling concerns about pricing models. Those who have the
responsibility to secure high-value mainframe environments will decide
just how important.

I have been a consultant to RSA Security for many years, and my bias is
obvious and acknowledged.

Suerte,

_Vin

------------------------------------------------------------
   Vin McLellan + The Privacy Guild + <[EMAIL PROTECTED]>
         22 Beacon St., Chelsea, MA 02150-2672 USA

Reply via email to