Wireless Broadband Caching Pages

Shades799

Hi All,

Just wondering has anyone else noticed any of the wireless providers displaying cached pages rather than the "live" page.

On a couple of instances I have used both 3 Wireless Broadband and Vodafone 3G to get onto pages and I have been displayed old versions of pages and sometimes logged into another persons profile!!!

Anyone else experiencing issues like these??

aidan_walsh

Depending on the sites architecture, it may be that they still have sessions open against the IP that you are assigned when you connect. Probably unlikely, but possible.

Shades799

Would I be engaging in "divilment" to suggest that wireless providers like this would cache pages in an attempt to increase the browsing speed?

paulm17781

I've never used the 3G services but I recall people here saying that "3" had you behind a proxy server. This would explain caching and other people's sessions coming up. I have no proof this is the case but it is what I heard some time ago.

ArseBurger

It's called transparent caching/proxying.

ISP's have used proxy caches for years. All Irish dial up providers in the 90's had them. No real need with DSL nowadays though.

GPRS/3G/EDGE/HSPDA are the modern version of dial up. Hence their re-emergence.

PogMoThoin

Correction, neither 3 or Vodafone are Wireless Broadband. Change the thread title to "Mobile Internet caching pages", wireless broadband is a whole different ballgame.

smellslikeshoes

PogMoThoin wrote: »

Correction, neither 3 or Vodafone are Wireless Broadband. Change the thread title to "Mobile Internet caching pages", wireless broadband is a whole different ballgame.

Calling the 3G providers wireless broadband is perfectly apt. By definition they are broadband regardless of what people on this board call them to differentiate them from fixed wireless providers. Just because of the amount of users on the sector can in theory lower speeds to dialup doesn't stop them being broadband.

zilog_jones

BT apparently use (or at least sometimes use) a proxy cache for their ADSL - I occasionally got blank pages with a Squid-generated footer when they were having trouble with Google a while back.

Correction, neither 3 or Vodafone are Wireless Broadband. Change the thread title to "Mobile Internet caching pages", wireless broadband is a whole different ballgame.

I don't know what your problem is - HSDPA is commonly accepted as a broadband technology, and in comparison to GSM and GPRS it definately is... most of the time. "Wireless broadband" is not specific to a particular system.

ArseBurger

PogMoThoin wrote: »

Correction, neither 3 or Vodafone are Wireless Broadband. Change the thread title to "Mobile Internet caching pages", wireless broadband is a whole different ballgame.

GPRS/3G/EDGE/HSPDA are all channelised multi frequency data transmission mechanisms that are transported over the air. As technologies they are all 'broadband'.

So calling it wireless broadband is the most accurate description.

You don't state that a VW Polo isn't a car simply because it's not up to the same capabilities as a Porsche.

Onikage

ArseBurger wrote: »

You don't state that a VW Polo isn't a car simply because it's not up to the same capabilities as a Porsche.

What if it had two wheels?

But really, it is confusing and misleading to lump 3 or 4 very different technologies under the "wireless broadband" name. I spent about 5 minutes explaining the difference to a PhD the other day and his opinion was that it is a "stupid name".

Shades799

ArseBurger wrote: »

It's called transparent caching/proxying.

ISP's have used proxy caches for years. All Irish dial up providers in the 90's had them. No real need with DSL nowadays though.

GPRS/3G/EDGE/HSPDA are the modern version of dial up. Hence their re-emergence.

To get back to the original discussion it appears that this is the case. It appears that that regular html/htm pages are cached while dynamic content and pages behind SSL are not.

I suppose its a legitimate way of improving this "mobile wireless broadband" but it is definitely something that modern day web programmers will need to take into account. I would expect programmers that have been in the game since the dial-up era are probably handling this without even thinking about it.

watty

No
GPRS is narrowband. Only dialup speed

Typically 3G/EDGE are only twice ISDN 2ch and worse latency.

HSDPA is not stable. It varies from 70k to 3000k. (That's only sometimes Broadband).
HSDPA latency is 120ms best (worst Broadband) to over 1000ms (worse than Satellite).

The OECD and EU only count FIXED external aerial links of 3G/HSDPA deployed as an alternative to DSL as Broadband. They do not count GPRS/EDGE/3G/HSDPA mobile as broadband at all. You can't control contention on Mobile Internet.

Wireless Nomadic solutions similar to Ripwave/Clearwire are similar or worse performance to HSDPA and are not counted by OECD as Broadband. You can't ensure a minimum broadband performance or control contention. Many of these Nomadic Technologies originally developed as ISDN wireless local loop replacements for rural areas. They've stretched 64k tech to 220k.

Fixed wireless solutions have controlled contention, lower latency and minimum performance speed of many times ISDN. They vary from a bit poorer than entry level DSL to x10 performance and lower latency of entry level DSL depending on technology.

The 3G operators should be banned from calling it Broadband. It's Mobile Internet.

Mobile WiMax and Mobile Flash-OFDM are in a greyer area between HSDPA (mobile Internet) and DSL (Broadband) performance as they can be poorer than entry level DSL, but for maybe 50% of users or 50% of locations are better than entry level DSL.

Mobile WiMax on 2.6GHz or 3.5GHz as envisaged in Ireland is NOT Mobile Broadband as the frequency is too high for mobile coverage (those are near LOS fixed Wireless bands in reality).

So far in Europe the Mobile Flash-OFDM is 450MHz or 870MHz, which does give true Mobility. 3G/HSDPA is 2100MHz which is a problem for coverage compared to GPRS/EDGE on 900MHz or 1800MHz. Obviously the best mobility in Mobile Phone Operator GPRS/EDGE/3G/HSDPA is the 70k GPRS or 240Kbps EDGE on 900MHz.

The biggest issue for EDGE/3G/HSDPA is cell sector usage. HSDPA / 3G is particularly poor on scaling wasting over 1/2 the capacity as sector fills, so a 3600Kbps sector with 20 users can easily drop to 70kps per user or less rather than linear 180kbps. The 25th user can't even connect.

I'd not call that Broadband, the EU & OECD doesn't either.

In fact it is a fast flat rate dialup. You can use Dialup networking instead of supplied GUI. Drop a connection and you lose the session (not true with "proper" broadband Wireless tech).

So, yes many dialup ISPs used/use proxies and even image degraders to speed pages. Why is it any surprise that GPRS/EDGE/3G/HSDPA does it?

ArseBurger

Onikage wrote: »

I spent about 5 minutes explaining the difference to a PhD the other day and his opinion was that it is a "stupid name".

Most PhD's opinions, in my opinion, are worth less than those of the common man on the street. All a PhD proves is that you've managed to avoid real work for an additional 3-5 years and can blinker yourself on a single topic.

I've interviewed a number of PhD's straight out of college. Not a single one had even grad level capabilites for Sys Admin.

ArseBurger

watty wrote: »

No
GPRS is narrowband. Only dialup speed

That's a fairly general comment considering you don't reference specific channelisation or frequency ranges...

And generally, it's wrong.

ArseBurger

watty wrote: »

HSDPA is not stable. It varies from 70k to 3000k. (That's only sometimes Broadband).

Broadband defined has little to do with throughput...

The OECD and EU

Not standards bodies.

Wireless Nomadic solutions similar to Ripwave/Clearwire are similar or worse performance to HSDPA and are not counted by OECD as Broadband.

The are broadband. freqs + channelisation again.

You can't ensure a minimum broadband performance or control contention.

Yes you can.

Many of these Nomadic Technologies originally developed as ISDN wireless local loop replacements for rural areas. They've stretched 64k tech to 220k.

Wrong. They're OFDM based. Not any relation to or based upon ISDN - unlike GSM/GPRS/3G/HSPDA/EDGE

The 3G operators should be banned from calling it Broadband. It's Mobile Internet.

Technical vs media definitions.

Mobile WiMax on 2.6GHz or 3.5GHz as envisaged in Ireland is NOT Mobile Broadband as the frequency is too high for mobile coverage (those are near LOS fixed Wireless bands in reality).

Incorrect.

So far in Europe the Mobile Flash-OFDM is 450MHz or 870MHz, which does give true Mobility.

Mobility and it's capability is not based upon frequency rather than coverage and RF fill.

The biggest issue for EDGE/3G/HSDPA is cell sector usage. HSDPA / 3G is particularly poor on scaling wasting over 1/2 the capacity as sector fills, so a 3600Kbps sector with 20 users can easily drop to 70kps per user or less rather than linear 180kbps. The 25th user can't even connect.

Then by your statement above, they are not mobile....

In fact it is a fast flat rate dialup. You can use Dialup networking instead of supplied GUI. Drop a connection and you lose the session (not true with "proper" broadband Wireless tech).

Completely depends upon the AAA mechanism behind the tech. Not relevant.

cgarvey

ArseBurger wrote: »

Broadband defined

Who/what do you use as your definition of Broadband?

ArseBurger

IEEE of course. Wouldn't think of using anything else for Telecoms.

smellslikeshoes

watty wrote: »

The 3G operators should be banned from calling it Broadband. It's Mobile Internet.

Mobile WiMax and Mobile Flash-OFDM are in a greyer area between HSDPA (mobile Internet) and DSL (Broadband)
performance as they can be poorer than entry level DSL, but for maybe 50% of users or 50% of locations are better than entry level DSL.

There is no grey area here, your argument that Hspda is not broadband because its possible to get to close to dialup speeds applies to Digiwebs Flash-OFDM the same as HSDPA. As I understand it Flash-OFDM doesn't degrade is speed as badly as HSPDA but it can still get down to ISDN style speeds or close to it. You have only to look at another thread on Digiewbs speed for proof of this. http://www.boards.ie/vbulletin/showthread.php?t=2055294712

We all know the problems that the number of users cause on HSPDA cells but saying that its not broadband because it can get close to dialup speed is not exactly fair when the same same can be said of fixed wireless or even DSL in some cases. I have seen plenty of cases of fixed wireless providers with speeds close to dialup at peak times, should these be stopped from advertising as broadband because of contention? Surely Watty you can't say with 100% certainty that a Digiweb Metro connection has never gotten down to ISDN speed because of contention?

As the technologies progress further and mobile broadband goes further into public knowledge the differences between fixed broadband and mobile broadband will become clearer and people will know the difference. You can't lump Hspda into the same category as dialup, when there are numerous happy customers getting 1 and 2 mbit regularly with Mobile Broadband.

Onikage

Shades799, naming conventions aside I have indeed noticed caching on the 3 network for some time now. At first I thought it was my browser compensating for the slow connection but then I found I was getting content older then the content in my cache, and it was definitely coming down the pipe (i.e. slowly)

I'm glad to say I haven't noticed other peoples sessions being served to me. If they were caching ssl pages that would be a cause for concern and possibly solicitors.

watty

No HSDPA is not dialup speed. But it's not accepted Internationally as Broadband either. However no implementation anywhere is other than Dialup networking type sessions. Session is lost on loss of signal or contention disconnect (voice calls have priority). Other newer "designed for IP" from the beginning Mobile Technologies do maintain the session during loss of connection up to a generous timeout limit.

By definition it's physically impossible to control contention on Mobile or Nomadic systems (in terms of number of users trying to connect in a sector). The Operators of Fixed Wireless can control contention by use of technology AND refusing further customers in a sector.

900MHz was picked for original GSM BECAUSE it is a good Mobile frequency. It's recognised that 3G's 2.1GHz is poor, hence intention of EU to reuse GSM 900 for 3G.

3.6GHz has been tested already in Ireland for Mobile and for practical purposes doesn't work. 2.6GHz (proposed future EU WiMax, current WiMax is 3.6GHz) is obviously closer to 3G's mobile performance than 3.6GHz Ripwave Nomadic performance.

OFDM is not to do with how mobile a system is or how fast, That is more Modulation, spectrum and frequency. OFDM allows better multipath (more relevent at 900MHz than 3600MHz) and better performance in presence of interference, thus allowing higher QAM / higher overall throughput. OFDM will give more advantage as frequency tends toward UHF and becomes a disadvantage as frequency tends to 10GHz.

Older systems are CDMA based because less terminal/Modem DSP is needed. Cheaper. But not very spectrum efficient (bits / hertz) and degrades badly in latency and speed as users added. many nomadic systems are CDMA based, even though they are field upgradable to OFDM Wimax or OFDM pre-wimax standards.

If I was claiming Flash-OFDM and Mobile WiMax always gave Broadband performance (I don't) I wouldn't class them as "Grey". I believe I mentioned hypothetical percentages.

AFAIK, eircom won't supply DSL if it is less than 512k and their entry level package is 1Mbps. I'm not sure there are any FIXED Wireless (not Nomadic) that regularly give less than 512K unless the CPE is broken or site is unsuitable. Generally WISPs refuse to install Fixed wireless nowadays unless the signal test and site suggest it will be reliable Broadband.

There is a big difference between inherent performance of Mobile technologies and comparing that with mis-installed or faulty Fixed Wireless or DSL or cable.

If GPRS was ever able to give anything remotely like BB performance (it's less than 2ch ISDN) why develop EDGE, 3G and HSDPA?

Marlow

Onikage wrote: »

I have indeed noticed caching on the 3 network for some time now. At first I thought it was my browser compensating for the slow connection but then I found I was getting content older then the content in my cache, and it was definitely coming down the pipe (i.e. slowly)

3 does use NAT firewalling and thus not serve public IP's to end-users and transparent proxy for sure. The proxies have often enough failed, so that people would have noticed.

Transparent proxies are often used to accelerate serving of content and preserve upstream bandwidth. In some cases they can improve the web-experience, if the provider that is using them has problems with latency, connections etc.

Many broadband providers choose not to use transparent proxies, because they are a point of failure and caching becomes pointless, because the amount of pages served over a day would require huge amounts of storage to make proxying effective.

/Martin

ArseBurger

watty wrote: »

By definition it's physically impossible to control contention on Mobile or Nomadic systems (in terms of number of users trying to connect in a sector).

Incorrect. But I'd love to hear why you think it's physically impossible by definition...

Also, I think you're confusing bearer and control plane traffic.

watty

You can't control where the modems are and how many there are.

On fixed Wireless you can. On fixed wireless 48:1 is a meaningful phrase. That is the the number of subscribers you will allow / sell on basis of 48 * sector capacity = Sum of all Subscriber packages. Since the system could be 100 or 200 users in a sector and all are not using at once, people rarely experience the 48:1 contention.

On Mobile/Nomadic the sector may only support 24 connections (3.6Mbps HSDPA). But the number of subscribers can't be controlled. You can't tell how may of your subscribers will be there as they are Mobile/Nomadic. So typically the full 24:1 (which gives typically 50:1 speeds or worse on CDMA based systems like HSDPA) is experienced much more often than even 3:1 contention on a Fixed Wireless system enginneered for 48:1 And of course on Fixed Wireless every subscriber is G'teed a connection and on HSDPA the 25th user can't connect.

Your flat contradictions with no proper explanations will just confuse people.


It's physically impossible to govern how many mobile/nomadic users attempt /desire to use a sector. Contention is meaningless. All you can describe is a max number of users and the performance you will get. This is not Contention in DSL, Satellite, Cable, Fibre or Fixed wireless sense where every subscriber (on properly engineered systems) will get a connection and there is a worst case engineered design contention.

A maximum number of connections on a system is NOT the same as contention. For example cable can support almost always more connections than the desired contention.

On Fixed Wireless Licences Comreg specifies a maximum 48:1. Some suppliers always have less than this and many systems can support higher contention.

The physical maximum number of connections on Fixed or Mobile/Nomadic is not the Contention ratio.

ArseBurger

watty wrote: »

You can't control where the modems are and how many there are.

Yes you can.

Your flat contradictions with no proper explanations will just confuse people.

Cell breathing.
Base busy notification.
Control plane negotiation.
Dynamic QAM scanning and attachment.

Then there's RF design and capacity engineering. Infill with micro and pico cells etc...


That's only at the RF side of things. You can do some funky stuff at the CSN side of things too to mange devices across your network.

Oh you know, like they do with GSM and stuff...

watty

Oh no you can't. That's what Mobile/Nomadic MEANS....

Look up Mobile and Nomadic in a Dictionary.

Yes cells breathe (A bad evil thing especially part of CDMA as extra handsets/modems reduce the SNR and thus terminals with poor SNR drop off)

Yes Dynamic QAM. It's irrelevent to number of connections and Contention. SNR again. With poor SNR (more range typically) you use QPSK. With strong signal you can increase from QPSK to QAM16 to QAM 64 for instance. Not Unique to any particular Mobile system and doesn't control the number of users that want to connect. Just improves speed for the particular connected user.


On a mobile/nomadic system the tecnology decides the maxiumum (typically 1/10th of fixed system) possible conncurrent connections. The Users, not network operators decide where the handset/modem is. It can be anywhere. Unlike Fixed wireless it may have no relationship to subscribers address. Anyway NO mobile operator says "Oh we have sold too many modems/handsets in Bray". Fixed Wireless operators will refuse further subscriptions on a mast sector that is full.

Which part of Nomadic / Mobile concept is confusing you?

ArseBurger

watty wrote: »

Oh no you can't. That's what Mobile/Nomadic MEANS....

Oh yes you can. I've seen it working in practive.

Which part of Nomadic / Mobile concept is confusing you?

Neither when working on real networks. Both when you present them incorrectly.

The 'concepts' of nomadic and mobile are irrelevant to the capability of providing and controlling contention against base stations.

Are you confusing control and bearer planes?