Sudden Rush to Push

Interestingly, there’s been some noticeable developments around push architectures just recently. Matt Tucker managed to create a minor blog buzz with his post “XMPP (a.k.a. Jabber) is the future for cloud services” where he introduces the XMPP in context of large-scale systems integration. Excerpt:

XMPP was invented for instant messaging and presence, and is the dominant open protocol in that space. Instant messaging? Yep, it turns out that all of the problems that had to be solved for instant messaging make the protocol perfect for cloud computing: […]

The comments are definitely worth reading as well.

That post was also discussed in ReadWriteWeb, “Could Instant Messaging (XMPP) Power the Future of Online Communication?”, and Bill de hÓra throws in his educated round-up. Both are very interesting reads. I liked about this note especially (from latter; original emphasis):

Atom and XMPP can go anywhere – it’s about reach, and simplicity, two criteria for innovative apps – this stuff will go over phones, web, desktops, tv, planes, trains and automobiles.

Effectively—also surprisingly and unexpectedly—this is a verbalization of our tagline. (Everywhere Services.) Thus, Xernel’s case has been built on top of the “XMPP architecture”; quoted, as the most exercised features are actually real-timeliness and bidirectionality, but XMPP furthermore brings in other essential features.

However, XMPP is just the infrastructure which we’re pretty much assuming given. Though, also we’re building yet another xmppd, but we’re not considering that as the most essential layer. (We’re just not happy with creating service clouds¹ based on current implementations.) Our primary interest is in how to wire the real-world services to the XMPP world.

The XMPP infrastructure is not there yet, but if you skimmed the references above you probably saw the strong signs like TiVo and Twitter. The query-response architecture is currently being given life-extension treatment with some hacks² whereas we’re expecting a bit more fundamental paradigm shift starting to happen.

Small, but definite signs.

1) Please note the fundamental difference between providing vs. managing services via XMPP.

2) Surely, the hacks are better than nothing, but these patches are usually leaky and not viable in a longer term. In this case the long-living TCP connections are one problem in context of HTTP.