From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-oi0-x243.google.com (mail-oi0-x243.google.com [IPv6:2607:f8b0:4003:c06::243]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "smtp.gmail.com", Issuer "Google Internet Authority G2" (verified OK)) by huchra.bufferbloat.net (Postfix) with ESMTPS id D647F21F4F2 for ; Thu, 15 Oct 2015 18:59:08 -0700 (PDT) Received: by oiho185 with SMTP id o185so5251039oih.0 for ; Thu, 15 Oct 2015 18:59:07 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:date:message-id:subject:from:to:cc:content-type; bh=gnfuuSzk8VScNIwZnLKkz7BOfqyN3Pnf2CsDFSBDusI=; b=y3TlBnc3GAz5p2QAcpRi+Amn7IQlB/9s3Lke2MYiboXfjKjd0cBEo7PVNhuCkhMtRL DH9qhBvYrekZTW411sIkqMLcdpzKKFBicCyhb6s+B5UH1Kuq9aEtCM/seDVuP0ff8pD8 0NULCErLJPbpTLY6labxescpi/9KEeW1NhhdjpaeGDYzEZM3zU7NdjkE6mF2L/4uIAyI vYi3p2zIRJOOaRNo5Xr3mXE4NXBEFuelSdR5XvsGhM+UTi0DLeusAmw9uhQTUA9DbuBD 9oF8F8GwM4sA9l0GSgEOVjI9+xJFtb4/eZVw1x9ddv4xK1KSCv5eOqPKCWv5vsjp5O1+ 3X1Q== MIME-Version: 1.0 X-Received: by 10.202.203.74 with SMTP id b71mr3442759oig.104.1444960747723; Thu, 15 Oct 2015 18:59:07 -0700 (PDT) Received: by 10.202.108.212 with HTTP; Thu, 15 Oct 2015 18:59:07 -0700 (PDT) Date: Fri, 16 Oct 2015 03:59:07 +0200 Message-ID: From: Dave Taht To: bufferbloat-fcc-signers@lists.redbarn.org, make-wifi-fast@lists.bufferbloat.net, Battle of the Mesh Mailing List , fcc@lists.prplfoundation.org Content-Type: text/plain; charset=UTF-8 Cc: Bill Weinberg Subject: [Make-wifi-fast] making wifi fast - analysis and ideas X-BeenThere: make-wifi-fast@lists.bufferbloat.net X-Mailman-Version: 2.1.13 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 16 Oct 2015 01:59:31 -0000 Below is some older analysis by a marketing expert of the make-wifi-fast project plan (although it is linked to in fcc wifi letter, it is so long that I doubt few as read as far as the detailed - and ongoing!!! - problems in wifi) the planned - OPEN - engineering work on wifi is buried in the In closing section - here: https://docs.google.com/document/d/1Se36svYE1Uzpppe1HWnEyat_sAGghB3kE285LElJBW4/edit We are always looking for contributions of funds, resources, and engineering time towards actually making wifi better. I had hoped to kick off the per-station queuing portion of the effort last week, actually. Is there anyone here that is actually a member of the wifi alliance? Has anyone reached out to them? http://www.wi-fi.org/who-we-are/member-companies There are quite a large number of companies that have got their seven thousand dollar coffee cup from the alliance, after passing their lousy tests for certification. ... snip snip - analysis by bill weinberg of the make-wifi-fast plan .... Here are my main impressions of the material you supplied and the efforts required: DIVERSE / DIFFUSE ACTION REQUIRED About 1/3 fto 1/2 of the concrete actions required/recommended can occur "organically", that is, through ad hoc and semi-organized efforts of individuals and small teams. The remaining work will require fairly dramatic engineering investments and also cross-industry changes in mindsets and practices - not something IMHO achievable without a paradigm shift (see below) BIGGEST CHALLENGES - binary blobs and open behavior: Even after 20 years of open source driving ecosystem expansion, key players remain incredibly guarded with regard to "their high-value IP" (herein intellectual property, not Internet Protocol), all the way down to low-level code in networking drivers. In this case I speak from very direct experience, in particular with semiconductor suppliers who service the WiFi marketplace. They are the real source of binary blobs, or more specifically, obfuscation and lack of access to chipset particulars and the code that should enable maximum wireless silicon ROI. These companies, for better or worse, are IP-driven and almost always IP-paranoid, even in / especially around otherwise open source settings. It will require a simultaneous carrot and stick approach to get them to abolish the blobs: a clear value proposition to fully open source drivers and exposed MAC and performance assist technologies, and an unpalatable onus for continuing business-as-usual - legacy / backward wireless standard interoperability: across the spectrum of IT we see how backward compatibility provides short-term business gain and long-term ecosystem pain. My favorite examples include the x86 architecture, the ARM architecture, HTML attributes, legacy encryption standards (now enabling back-off exploits), the US and IPv4 and support for 16 and 32 bit infrastructure in an increasingly 64 bit world. The WiFi document highlights how backward compatibility is killing WiFi performance. My suggestion is the establish a new "basement", wherein devices after a certain point will only support newer standards, with explicit provision on how to handle legacy with parallel tech, as was the case in the early days of WiFi (e.g., with USB-based and SD-interface devices vs. native ones) - Co-opting the WiFi Alliance: the WiFi brand is currently owned by the WiFi Alliance who, IMHO, think that they are doing a great job of managing the brand and of promoting WiFi evolution. Not. BIG SOLUTION I think the key is to "invent" a new WiFi. It might really use the most advanced versions of existing tech, but it would get a new name, and most importantly, new standards that encompass many/most of the micro-solutions outlined in Jim Getty's "Make WiFi Fast" document, as follows: - a clean transition to the most advanced, forward-looking spec cum standard - a consortium to (re)brand the new medium (call it Firefly or what you will) and establish ground rules for participation, enforced by branding rights: * no binary blobs ~ fully OSS drivers and other support code, up to a pre-defined level in the stack * driver coding standards for Linux, Windows and RTOSes to promote openness, anti-buffer bloat, etc. * specific provision for the use cases mentioned in Jim's document (e.g., multicast, mesh, IoT, etc. [pardon non orthogonal members]) * specific rules on how to continue supporting legacy 802.11* with completely parallel infrastructure (albeit in same sorry band) * two (no less) strong founding members, comprised of one wireless chipset provider (of substance) and one OEM (ideal would be Intel and Cisco - dream on) and incentive for the rest of the gang to join on day 1 or 2 * special extraordinary efforts to get Asia on board - China in particular - to participate (concessions probably required, esp. around security, as before) * start with a professional body management company (e.g., Global Inventures) but envision a small professional staff, especially to manage endowments and to instigate marketing * such a new group would need to 1) subsume the WiFi Alliance, 2) co-opt it and effectively take its place, or 3) find a way to complement the WiFi Alliance. The alternative is to stage a putsch and take over the WiFi Alliance from within, probably on the backs of key sponsoring members (see http://www.wi-fi.org/who-we-are/member-companies) (am I overstating the actual importance of WFA?) ... sincerely Dave Taht http://www.bufferbloat.net/projects/bloat/wiki/Daves_Media_Guidance