doing simulations of xcache at US scale. Due to wide interest in doing the same for different sites, I will be giving hands-on tutorial on it during the next Friday ADC analytics meeting.
In essence, it seems that a two layer xcache setup would be sufficient. Now investigating differences between analy and prod queues data accesses.
Working out a plan to deploy Tensorflow as a service. Options are:
* TFAAS - from Valentin K.
* TF serving - from Google
* TFjs
Also looking at spark with TF