My private Electrum server is having trouble keeping up with the present network activity,  being up to 11 blocks behind.  The bitcoin node on the same machine has no trouble keeping up and the CPU time used by electrum-server is now 7 times greater than that used by bitcoind.

Worse, when I use an Electrum client to access the node I observe that the client reports an erroneous number of blocks lagging (more blocks than actual).  In addition, it does not pull out a new unconfirmed received transactions that have been in the mempool for over an hour.  (Switching the client to another server does show this unconfirmed transaction.)  This is a private server and only one client is accessing it.  Perhaps a better balance in thread priority can be taken between resources spent tracking the network and updating the database vs. providing timely interaction to clients.

Looking at the logs, it looks like one block gave the server serious cramps:

[08/07/2015-12:24:49] utxo not in database; postponing mempool update
[08/07/2015-12:24:49] blockchain: 364428 (5284.813s)
[08/07/2015-12:25:15] mempool_update 16.0112478733

The machine has a Haswell Atom processor, but it has 8 GB of RAM and an SSD.  Running Ubuntu 14.04, Electrum Server updated from github yesterday.