http://storm-project.net/

• Integration points would appear to be through sprouts (a source) and bolts (a sink). You can define/control the sprouts and bolts you use, so these would likely be the first integration points with SY.
• A JMS implementation is already provided for sprouts and bolts, so one could play around with initial integration by using JMS to bridge between SY and Storm.
• Defining a SY aware sprout and bolt implementation has some interesting possibilities - it would simplify the topology definition, wouldn't require specification of distinct gateway bindings (e.g. JMS) for SY services called from Storm, etc.
• The topology defines the execution sequence in a way a lot like a Camel route or action processing pipeline might. Might make for an interesting composition option if you could produce a deployable topology (all-in-one jar) with the appropriate SY dependencies and service implementations in place. This would be distributed out to all the execution nodes and the execution flow would be managed by Storm. Reaching a bit on this one, but it seems possible.