Event-Driven Programming Vivek Pai Dec 5, 2002

advertisement
Event-Driven Programming
Vivek Pai
Dec 5, 2002
GedankenBits
 What
does a raw bit cost?
 IDE
 40GB:
$100
 120GB: $180
USB Pen: $38
 FireWire:
 What
cost?
 NAS
 320GB:
$3400 (Dell)
 7.5TB: $300K (Zambeel)
 32MB
 60GB:
$205
 120GB: $280
 USB:
does a managed bit
 What
does a person cost?
 $100K
salary + $50K
benefits/overhead
 40GB:
$140
 120GB: $218
2
Extra Project
We
need to agree on it
One-page proposal
Idea, implementation, measurement
Report (~10 pages) due on Dean’s Date
Max boost of 2/3 of a letter grade
3
Project 6
Several goals
Performance
improvement via caching
Dynamic adjustment to load
Master/slave or symmetric programming
Due
on Dean’s Date
Extra credit: 4 points (20 base)
4
Another Random Aside
You
may want to read
Flash: An
Efficient and Portable Web Server
Available from my home page
Caveat: far more
complicated than Project 5
5
Official Goals
Discuss the
difference between standard
programming styles and event-driven
programming
Show the difference between structuring servers
using processes and application-level
multiplexing
Discuss the benefits and drawbacks of each
approach
6
What Is An Event?
Some
kind of notification
Interrupts
Signals
Polling
(via poll/select/etc)
Callback (via function pointer)
Similarities?
7
“Reactive” Environments
Windowing systems
Network-aware programs
Drivers of
all sorts
8
Traditional Environments
One
thing going on at a time
Finish that thing, go on to next
Any step can block indefinitely
Resumption from blocking is simple – OS
provided
9
What Goes On In A Web Browser?
Drawing
the current page
Inline images
Translating
machine names to IP addresses
Launching connections
Sending requests
Getting piecemeal responses, drawing images
User
clicking to next link
10
Threads Versus Events
One
stack versus many stacks
What happens on blocking operations
Parallelism
Shared variables
State
11
Let’s Think Of States
How
many possible states are there?
Take
all pieces of information
Decide valid range for all pieces
Enumerate
Can
we reduce states?
Some
combinations invalid
Still, lots of states
12
Really Reducing States
Take all
major pieces of program
Add extra tags to state
What do tags look like?
Position
Count
#
13
State Examples
If-then-else
3
For
1
states: start, then-clause, else-clause
loop
state + count
Why do
we care?
Resuming
at the right state
14
Remember This Diagram?
Accept
Conn
Read
Request
Find
File
Send
Header
Read File
Send Data
end
15
Structure of Event-Driven
Programs
Lots of
state machines
Maintaining information about each one
Some way of moving through states
As few restrictions as possible on timing
16
The Real Structure
While (1)
Get event
Dispatch event
Or,
while loop in library
Event
handlers in main program
17
Delays
Are
delays possible?
Interrupt
handlers – generally not
Otherwise? Depends on event rate
How
to avoid delays?
More
events – asynchronous operations
What happens if no async support?
 Fake
it
18
Select( ) System Call
int select(int nfds, fd_set *readfds,
fd_set *writefds, fd_set *exceptfds,
struct timeval *timeout);
FD_SET(fd, &fdset);
FD_CLR(fd, &fdset);
FD_ISSET(fd, &fdset);
FD_ZERO(&fdset);
19
20
Select Description
DESCRIPTION
Select() examines the I/O descriptor sets whose addresses are passed in readfds,
writefds, and exceptfds to see if some of their descriptors are ready for reading, are
ready for writing, or have an exceptional condition pending, respectively. The only
exceptional condition detectable is out-of-band data received on a socket. The first nfds
descriptors are checked in each set; i.e., the descriptors from 0 through nfds-1 in the
descriptor sets are examined. On return, select() replaces the given descriptor sets with
subsets consisting of those descriptors that are ready for the requested operation.
Select() returns the total number of ready descriptors in all the sets.
The descriptor sets are stored as bit fields in arrays of integers. The following macros
are provided for manipulating such descriptor sets:
FD_ZERO(&fdset) initializes a descriptor set fdset to the null set.
FD_SET(fd, &fdset) includes a particular descriptor fd in fdset.
FD_CLR(fd, &fdset) removes fd from fdset. FD_ISSET(fd, &fdset) is non- zero if fd is
a member of fdset, zero otherwise. The behavior of these macros is undefined if a
descriptor value is less than zero or greater than or equal to FD_SETSIZE, which is
normally at least equal to the maximum number of descriptors supported by the system.21
Blocking Steps
Disk Blocking
Accept
Conn
Read
Request
Find
File
Send
Header
Read File
Send Data
end
Network Blocking
22
Overcoming Disk Blocking States
Accept
Conn
Read
Request
Find
File
Helper
Send
Header
Read File
Send Data
end
Helper
23
New Architecture - AMPED
Asymmetric Multiple Process Event Driven
Accept
Conn
Read
Request
Find
File
Send
Header
Read File
Send Data
Event Dispatcher
Helper 1
Helper 2
Helper N
Helpers are threads or processes
24
Caches in Flash Web Server
Accept
Conn
Read
Request
Find
File
Send
Header
Read File
Send Data
end
Pathname Response Mapped
Translation Header
File
Cache
Cache
Cache
Helper
Helper
25
Download