RE: iSCSI CONNECT message

[Joshua Tseng/Nishan Systems <joshua.tseng@NishanSystems.com>]

Jim,

I think we're getting our definitions straight, but I am still
not sure which model we're talking about (implicit or explicit).
If the iSCSI initiator identifies a target using an IP address,
placing that IP address after the target: text in the login message,
then a storage controller must exist at that IP address and there is
no proxy or tunneling operation.  On the other hand, if the iSCSI
initiator identifies a target using a DNS name, it could be talking
to a proxy OR the actual storage controller, and it does not need
to know is the case.  If you're saying it is "explicit tunneling"
when the DNS name has to be used and not the IP address, then okay,
this is explicit tunneling.

However, do note that if the iSCSI initiator identifies
a target with the DNS name, it should not care whether it is talking
to a proxy or the actual storage controller.  Unless I've overlooked
something, this should be completely transparent and non-material
to the iSCSI initiator, who shouldn't care either way.  If a proxy
resides at that IP address, he will understand the iSCSI login, set
up the appropriate TCP state tables, do its own DNS lookup, and
forward login the message on to the next proxy, or to the actual
storage controller.  But the original initiator shouldn't need
to be aware of this operation by the proxy.

I am neutral as to whether a CONNECT message should be included
in iSCSI for architecturual purity reasons, but I agree that
technically, the login message alone has enough information to
provide connectivity through proxies.  If we use a CONNECT: message,
then we don't need the DNS name in the login message.  I guess the
question is whether CONNECT and login should be done in the same
message.

Josh

-----Original Message-----
From: Jim Hafner/Almaden/IBM [mailto:hafner@almaden.ibm.com]
Sent: Monday, October 09, 2000 10:14 AM
To: ips@ece.cmu.edu
Subject: RE: iSCSI CONNECT message


Josh,

OK, I think we're getting to the crux of the matter (and getting our
definitions straight).

Implicit or transparent tunnelling:  NO *protocol specific* message is
needed to TCP-connect an iSCSI initiator to an iSCSI target.  (E.g., this
is the case, if an intermediary (proxy) aliases an ipaddress:port for a
target).

Explicit tunnelling: a protocol specific message, handled by the
intermediary, is required in order that a TCP connection be established
between an iSCSI initiator and iSCSI target (in those cases where the
implicit tunnelling is not sufficient).

I've heard both sides argued.  In my opinion, more people believe in the
second (Josh, Julo, ?).

Among those that believe the second is "where it's at", there is some
disagreement concerning what that message should be.  There are two
proposals:
(a) CONNECT message.
(b) iSCSI Login message with "Target:..." text field.

The iSCSI login message with "Target: ..." text is an "embedded" explicit
tunnel instruction AND much more (as it has context that is relevant only
to the iSCSI initiator and iSCSI target, not to the intermediary reading
the tunnel instruction).   In this case, the login message is carrying two
different kinds of information (one for the intermediary and one for the
iSCSI target at the end of the pipe).   I consider the iSCSI Login as
establishing context for the iSCSI initiator and the iSCSI target and NOT
for any intermediary.  [Maybe I'm wrong here!]

IMHO, the CONNECT message is nothing more or less than the explicit tunnel
instruction, and as such is the cleaner approach.   It is easy for an
intermediary to handle; it has no extraneous information; it's layered
bettered.

OK, now I'll try to shutup on this point.   If the WG thinks that iSCSI
login should provide this embedded tunnel instruction, so be it (but I
disagree!).

Jim Hafner


Joshua Tseng <jtseng@NishanSystems.com>@ece.cmu.edu on 10-09-2000 09:17:49
AM

Sent by:  owner-ips@ece.cmu.edu


To:   ips@ece.cmu.edu
cc:
Subject:  RE: iSCSI CONNECT message



Jim,

Whether we are talking about implicit or explicit tunneling depends
on your perspective.

>At the end of the note you say that my condition (A) "an initiator can
>ALWAYS open a connection to a target through normal TCP/IP mechanisms" is
>false. That is, we need explicit tunnelling.

I define TCP/IP mechanisms as IP address and port number (not DNS name).
Therefore, from the TCP/IP perspective, the target's IP address and port
number may not be visible to the initiator (although the DNS name is).
This is what you call explicit tunneling, if you are talking about TCP/IP
(address & port).

>Perhaps my choice of words is misleading.  By "open a connection", I did
>not mean "direct" connection but a connection (perhaps through multiple
>gateways/tunnels, etc.) that at least as far as the host "opensocket" call
>is concerned is transparent.

This is true.  From the perspective of the initiator host, the
tunneling is transparent.  The initiator host knows the target by
the DNS name, not the IP address & port.  The proxy is responsible
for the tunneling. operations.  I think this is what you mean by
implicit tunneling.

Josh

-----Original Message-----
From: Jim Hafner/Almaden/IBM [mailto:hafner@almaden.ibm.com]
Sent: Monday, October 09, 2000 8:33 AM
To: Joshua Tseng
Cc: ips@ece.cmu.edu
Subject: RE: iSCSI CONNECT message



Joshua,

I'm confused.

At the end of the note you say that my condition (A) "an initiator can
ALWAYS open a connection to a target through normal TCP/IP mechanisms" is
false. That is, we need explicit tunnelling.

Then in the beginning of the note you say that  DNS can get me deep into
the bowels of a private network.   To me that says that the world is
"transparent tunnelling".
Perhaps my choice of words is misleading.  By "open a connection", I did
not mean "direct" connection but a connection (perhaps through multiple
gateways/tunnels, etc.) that at least as far as the host "opensocket" call
is concerned is transparent.

Again, I'll state my case:
If we need explicit tunnelling (like "Target: ..." string in login), then I
contend the CONNECT message is better suited to that purpose.
If we don't need explicit tunnelling, then drop my CONNECT proposal (and
"Target"... string in login!).

Jim Hafner


Joshua Tseng <jtseng@NishanSystems.com>@ece.cmu.edu on 10-07-2000 09:52:47
AM

Sent by:  owner-ips@ece.cmu.edu


To:   ips@ece.cmu.edu
cc:
Subject:  RE: iSCSI CONNECT message



Hi Jim,

<snip..snip>
>Use of DNS: there may be security concerns here (about DNS itself).  But
>this also assumes that every iSCSI target has a "public" ipname (or
perhaps
>ipname:port combo).This may or may not be the case (correct?) if the
>controller lives deep in the bowels of some private network.

If the controller is in the bowels of some private network, it should still
be addressable by DNS, as long as the root authority is talking to the
root ICANN servers.  Hence, the following DNS name:

     disk4.hpnetworkA3D.hpnetworkA3.hpnetworkA.hp.com

is resolvable on the public Internet as long as the DNS server for
"hp.com" is talking to the ICANN ".com" server, and the DNS server
for hpnetworkA.hp.com is talking to the server for hp.com, and....

Security concerns about DNS can be handled separately through independent
authentication and/or encryption mechanisms between iSCSI entities and/or
proxies.

<snip..snip>
>the pipe is open for them to talk to each other).  iSCSI security may be
>completely independent of the link security (e.g., that the gateway might
>want to impose).  The iSCSI login security involves a context that is only
>relevant to the two end points as iSCSI entities, not as TCP/IP entities
>(i.e., at a different layer).  The link security is potentially
independent
>from the iSCSI security context and is a function of the two ends of an
>intermediary link (as TCP or IP entities).

Jim, I am in complete agreement here.  I would like to add that IPSec
provides
security between IP endpoints.  IPSec provides network level security,
while
TLS or iSCSI security can provide security for iSCSI entities, since
SSL, TLS, and iSCSI security only protects the TCP payload (or a subset of
the payload) and not the IP or TCP header, so it can be proxied without
changing the payload.  IPSec protects the IP header, so it can't be
proxied.
Rather, the proxy must authenticate and/or decrypt the IPSec before it can
forward the data to the next IP endpoint.

    network domain 1  |   network domain 2   |    network domain 3
                      |                      |
iSCSI initiator-----proxy1-----------------proxy2-----------iSCSI target
      |               |                      |                  |
      |<---IPSec----->|<-------IPSec-------->|<-----IPSec------>|
      |<---TCP 1----->|<-------TCP 2-------->|<-----TCP 3------>|
      |                                                         |
      |<-------------------iSCSI security or SSL/TLS----------->|
      |<-------------------iSCSI session----------------------->|
      |                                                         |
      |<--------------------SCSI session----------------------->|

I believe this security model is quite practical as well, since there
is no dependency between IPSec and iSCSI.  If the administrator wants
to protect the proxys, then IPSec can be added and the iSCSI layer and
your CONNECT mechanism will be completely ignorant of the presence or
nonpresence of IPSec (IPSec has its own key distribution mechanism).
IPSec can be managed separately and independently.

<snip..snip>
>In short, I think I can summarize the issues:
>A) if an initiator can ALWAYS open a connection to a target through normal
>TCP/IP mechanisms, then there is no need for my proposal. (I didn't think
>this was necessarily possible).  Additionally, this assumption implies
that
>target naming is pure and simply an ipname:port and nothing more (that is,
>I don't need URLs or any other complicated naming scheme).
>B) if NOT, then my proposal defines a means whereby that initiatial
>connection can get established, in order that the rest of the iSCSI
process
>can begin. I think you need a two-part naming mechanism in this case.  If
>one was enough, then option A holds.

My experience with the Public Internet and corporate WANs says that A is
not true.  Sure, there will always be cases in a private network where the
administrator uses IP addresses only with no NAT, and is completely cut off
from the Public Internet (military/national defense concerns come to mind).
But if iSCSI is to be used through the Public Internet, then I believe
your B) might be the case.

Josh

-----Original Message-----
From: Jim Hafner/Almaden/IBM [mailto:hafner@almaden.ibm.com]
Sent: Thursday, October 05, 2000 9:33 AM
To: ips@ece.cmu.edu
Subject: Re: iSCSI CONNECT message


Folks,

Let's see if I can handle a bunch of these questions at once.  I'll admit
upfront that I'm not the most knowledgeable about how the IP network works,
how DNS works, how tunnelling works, etc.  As a consequence, I may be using
terms well-known in the network community in the wrong way.  Please correct
me if I am.

Definition: I'm using the term gateway here to mean any device (proxy,
etc.) with the following properites:
1) it sits between an initiator and a target (an implementation of a proxy
or any other sort of firewall)
2) it obscures the ipname/address of the target on its back-side from the
initiator on the front-side.
3) it is NOT an iSCSI target device; it is a device that enables connecting
two iSCSI devices
(in effect, a gateway is a device that must provide some sort of
tunnelling).  Or is "intermediary" a better term here?

Tunneling:  As Costa said, the only standardized tunnelling mechanism
defined (AFAIK) is in specific protocols like the HTTP GET URL protocol.
As I mentioned in my note, I'm suggesting that perhaps an analogous
function is required here.

Use of DNS: there may be security concerns here (about DNS itself).  But
this also assumes that every iSCSI target has a "public" ipname (or perhaps
ipname:port combo).This may or may not be the case (correct?) if the
controller lives deep in the bowels of some private network.

If the controller has a public IPname, then the normal mechanisms for
connecting to it should work (even through gateways as described by
Joshua).  In my proposal, the CONNECT message effectively gets delivered
directly to the target in the first step.

Is this the same as the login?  To me, the login is an initiator to target
operation, to validate the iSCSI to iSCSI layer connection end-to-end (once
the pipe is open for them to talk to each other).  iSCSI security may be
completely independent of the link security (e.g., that the gateway might
want to impose).  The iSCSI login security involves a context that is only
relevant to the two end points as iSCSI entities, not as TCP/IP entities
(i.e., at a different layer).  The link security is potentially independent
from the iSCSI security context and is a function of the two ends of an
intermediary link (as TCP or IP entities).  The CONNECT message then is the
instruction to the intermediary to request it's tunnelling services.
This gets to one of David's concerns about tunnel autoconfig.  My third
option (my favorite) for security in the CONNECT was effectivly leveraging
whatever tunneling autoconfig policies are in place between the two
endpoints of a hop (in the picture below, G1 and G2 may have their own
policies, which I assume they impose on each other, independent, perhaps,
of the type of traffic).

Julo's Topology(a):  I---G1---G2---G3---T

This is exactly the topology that Daniel and I discussed and the CONNECT
message was supposed to enable.  If this is "of little interest", then I
don't see the point of the CONNECT, either.  It may be that a gateway is
just a passthru or a proxy or any other mechanism that the gateway utilizes
in order to provide the services (QoS, security, etc.) that motivated the
placement of that gateway in that spot in the first place!

David also mentioned an issue about QoS and such.  If I'm a gateway doing
all this obsuring, then perhaps I'd like to have policies for QoS as well.
Whether they are blind to the type of traffic (iSCSI or http or ...), is a
different issue.

In short, I think I can summarize the issues:
A) if an initiator can ALWAYS open a connection to a target through normal
TCP/IP mechanisms, then there is no need for my proposal. (I didn't think
this was necessarily possible).  Additionally, this assumption implies that
target naming is pure and simply an ipname:port and nothing more (that is,
I don't need URLs or any other complicated naming scheme).
B) if NOT, then my proposal defines a means whereby that initiatial
connection can get established, in order that the rest of the iSCSI process
can begin. I think you need a two-part naming mechanism in this case.  If
one was enough, then option A holds.

Did I miss anybody's questions?  Am I completely off base here?  Can
somebody say whether (A) holds?  Does (A) hold with the requisite security
requirements (or is that a separate issue)?

Jim Hafner