Are we ever going to see External Storage for 622?

[John Kotches]

OTA HD is about 19mbps and the highest E* HD is much less. Lets say that each HD stream requires 25mbps throughput x 5 at once = 125mbps. USB2 is 480Mbps so it would barely tax the system. Bring it ON.:

]
Ok, the USB2 interface can theoretically sustain 480 Mbits/second...

Now go find me a hard drive that can sustain 125 Mbits/second at what would essentially be random writing of three consecutive streams. 125 Mbits / second is ~16 MB/second which is quite substantial when we're talking about sustained (not burst) I/O from a single spindle.

The 3 HD streams simultaneously is about the max you can get from a single 7200 RPM spindle and have margin for error.

Cheers,

 

[John Kotches]

Yeah I was pretty surprised that the 622 did not have an eSATA port, seems kind of elementary, 3.0Gbps vs the crappy 4xxMbps offered by USB. the real question is when can we get a Series 3 TiVo for Dish, then we can expand at will, there are folks on D* that have 1TB TiVo units!!!!

Again, the spindle will be the limiting factor, not the interface.

So the eSATA port would be you no real world performance benefits over USB2.

Cheers,

 

[navychop]

]
...The 3 HD streams simultaneously is about the max you can get from a single 7200 RPM spindle and have margin for error...

So maybe external storage will have a limit of one stream at a time. That's certainly more than good enough- more than we have now.

Keep in mind, those 3 (or 5) streams are working with a hard disk today- the one built into the ViP622.

 

[John Kotches]

Navychop...

The 3 streams are each max about 20Mbits/second totalling an aggregate of ~60 Mbits/second. This is a far cry from the interface limits of 480Mbits/second (USB2) and SATA-II (3.0Gbits/second). This is a misunderstanding I was trying to dispel -- just because your interface supports a given data rate doesn't mean the device on the interface will max it out.

When talking about performance parameters it is important to understand the differences between short term burst transfers which are largely done from the drives internal cache and sustained sequential I/O and sustained random I/O. Those are listed in order of decreasing performance.

We're talking about A/V files which aren't going to be short term bursts so take that off the table. That's the only transfer that's going to come anywhere near the interface limit for USB2.

So we're left with sustained (non-random) I/O and sustained random I/O. For read and writes, you're going to fall somewhere in between these two values; leaning closer to the random side because of the seeks required to move from segments of data that comprise the already stored (recorded) or actively being written (recording in progress) file.

In a perfect world, the data files would be straight sequential I/O; but the real world isn't perfect and it's going to be segmented somewhat; depending on I/O load, amount and location of available space.

But what you will never see is sustained transfer approaching the interface speed of USB2 (let alone SATA-II) from a single 7200rpm or even 10K rpm spindle.

You can get close with larger RAID arrays; but those are bigger $$$ than most consumers are willing to spend.

Cheers,

 

[John Kotches]

Navychop:

Also, at one point Scott had mentioned that it was possible the external drive would be for archiving only; ie you would take programs from internal storage and transfer them to the external storage and that all recording would take place on the internal drive.

If that's the case, the argument could easily be made that 0 active recording streams are available on the external device

Cheers,

 

[navychop]

Yes, I understand that. But in response to
"...The 3 HD streams simultaneously is about the max you can get from a single 7200 RPM spindle and have margin for error...."
I am pointing out that today we do the job with an internal hard drive. And if the interface is not the limit, and the hard drive can do it internally- what's the problem with doing 1, 3 or even 5 streams externally? Or even go for the marginal improvements of a 10,000 RPM drive. The spindle doesn't seem to be limiting at all. Or am I missing something? Overhead is not likely to be over 1/3 of the total, is it?

 

[Jim5506]

Let's see, the 622 can record three HD streams while playing 2 - that's 5 -Right?

 

[Stargazer]

5 functions on 3 streams, not 5 streams.

 

[John Kotches]

Yes, I understand that. But in response to
"...The 3 HD streams simultaneously is about the max you can get from a single 7200 RPM spindle and have margin for error...."
I am pointing out that today we do the job with an internal hard drive. And if the interface is not the limit, and the hard drive can do it internally- what's the problem with doing 1, 3 or even 5 streams externally? Or even go for the marginal improvements of a 10,000 RPM drive. The spindle doesn't seem to be limiting at all. Or am I missing something? Overhead is not likely to be over 1/3 of the total, is it?

You're missing that the maximum is 100 Mbits/second which is really pushing your luck. 5 streams in this fashion of random I/O is potentially troublesome.

Try it some time and see what the results are. I'm not even sure you can have 5 simultaneous streams working from the hard drive.

Cheers,

 

[navychop]

Well, I'm in single mode, so 4 is my limit. Might try it sometime. Not while recording anything important.

 

[Julien]

Well, I'm in single mode, so 4 is my limit. Might try it sometime. Not while recording anything important.

Only if and when E* enables it.

 

[navychop]

Ah. The reference was to see how well the internal hard drive could handle a max load. I have recorded 2 streams while watching a third, but I do not recall ever recording 3 streams while watching a fourth.

I'm hoping we'll hear more next week. Hopefully not a statement with the dreaded word "soon" in it. Preferably something like "AVAILABLE NOW." And "for no additional charge."

 

[dfergie]

I'm hoping we'll hear more next week. Hopefully not a statement with the dreaded word "soon" in it. Preferably something like "AVAILABLE NOW." And "for no additional charge."

Yes... dish throw us a bone...

 

[Jim5506]

5 functions on 3 streams, not 5 streams.

The stream only flows one way?

 

[RTCDude]

Is this ever going to happen? Back in Feb 06 we were promised Fall. It's now Winter.

Brian

Try waiting for NBR on 5xx machines! It's been 2 and a half years now since it was "promised" (I think, or has it been longer?!?!?).

 

[John Kotches]

The stream only flows one way?

Well there are live feed streams and playback streams. Live feeds are write streams and playback streams are read streams.

Cheers,

 

[Smith P.]

in term of bytes per second there're the same

 

[John Kotches]

in term of bytes per second there're the same

Since you didn't quite a post you were responding to and it isn't clear...

In terms of USB usage, yes it is the same...

In terms of the mechanism read performance will always exceed write performance.

Best,

 

[Smith P.]

Only if using RAID-5 ! Tested.

 

[Tom Bombadil]

3 days until CES. Most announcements are made in the first day or two. But knowing E*, they will be late.