Marimba (Castanet)

» see marimba.com as it lived, on the wayback machine

~ the last words ~

in '96 i was twelve. our house in turkey did not have a modem yet. marimba was founded that february in mountain view by kim polese and three java engineers from sun who left to build push technology. polese had been the public face of java at sun, the woman on the cover of newsweek when the language launched. the marimba pitch was that the web's pull model would invert. servers would push content and software updates down to client machines on a schedule instead of waiting for the client to ask.

the product was called castanet. it sat between an installer and a content delivery network. publishers built channels. users subscribed. castanet handed updates from server to client in increments, only the diff, only when the channel was ready. companies like disney and fedex set up channels. wired put push tech on its cover in march '97. for a stretch of about eighteen months the silicon valley narrative was that the web as a pull medium was about to be replaced by something that worked more like television.

the push wave broke fast. by 1998 it was clear most users hated being pushed at. the bandwidth math did not work for consumer 56k modems either, and the editorial workload of building daily channels exceeded what any publisher was willing to staff. marimba pivoted to enterprise software distribution where the math worked because the network was a corporate lan and the client was an IT department's managed desktop. polese stayed CEO until '00. the company ran a quiet enterprise business until bmc bought it on the fifteenth of july 2004 for two hundred thirty nine million dollars. the consumer brand died at the acquisition.

~ the record ~

~ the push wave ~

the 1996-1998 push-tech moment is its own small chapter in web history. pointcast was the consumer demo. marimba was the platform pitch under it. a dozen smaller channels lived inside both. the assumption was that the desktop screensaver would become a kind of cable channel and that the daily newspaper was about to be replaced by a stream of small notifications. that assumption was wrong by 1999. the screensaver did not become cable. the newspaper did not get replaced for another twenty years, and the channel metaphor moved to rss before dying there too.

~ the pivot ~

marimba survived the bust by quietly turning castanet into an enterprise change-management product. honestly, the technology was the same. you needed to roll out a new version of a desktop application to ten thousand windows machines on a corporate network. castanet would do that without breaking machines that lost connectivity mid-update. boring and repeatable revenue, the right kind of work to take across a recession. bmc folded the product into its remedy line after the acquisition. the marimba brand stayed on a few sku names for a year or two then dissolved into bmc's catalog.

~ what got buried with it ~

the consumer side of push technology died completely. by 2026 the only thing that remained of it is the os-level notification permission prompt on every smartphone, which inherits the basic idea minus the channel pretense. polese went on to start other companies. she took board roles and became an early-stage investor on the bay area circuit. the three ex-sun engineers spread back through the industry. marimba's name lives on the bmc website as a legacy product page that nobody updates. the wired covers from '97 still come up when somebody searches "push technology" looking for old screenshots.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »