Develop Reference

Sending data to a database in size-limited chunks

I have a method which takes a parameter which is Partition enum. This method will be called by multiple background threads (15 max) around same time period by passing different value of partition. Here dataHoldersByPartition is a map of Partition and ConcurrentLinkedQueue<DataHolder>.
private final ImmutableMap<Partition, ConcurrentLinkedQueue<DataHolder>> dataHoldersByPartition;
//... some code to populate entry in `dataHoldersByPartition`
private void validateAndSend(final Partition partition) {
ConcurrentLinkedQueue<DataHolder> dataHolders = dataHoldersByPartition.get(partition);
Map<byte[], byte[]> clientKeyBytesAndProcessBytesHolder = new HashMap<>();
int totalSize = 0;
DataHolder dataHolder;
while ((dataHolder = dataHolders.poll()) != null) {
byte[] clientKeyBytes = dataHolder.getClientKey().getBytes(StandardCharsets.UTF_8);
if (clientKeyBytes.length > 255)
continue;
byte[] processBytes = dataHolder.getProcessBytes();
int clientKeyLength = clientKeyBytes.length;
int processBytesLength = processBytes.length;
int additionalLength = clientKeyLength + processBytesLength;
if (totalSize + additionalLength > 50000) {
Message message = new Message(clientKeyBytesAndProcessBytesHolder, partition);
// here size of `message.serialize()` byte array should always be less than 50k at all cost
sendToDatabase(message.getAddress(), message.serialize());
clientKeyBytesAndProcessBytesHolder = new HashMap<>();
totalSize = 0;
}
clientKeyBytesAndProcessBytesHolder.put(clientKeyBytes, processBytes);
totalSize += additionalLength;
}
// calling again with remaining values only if clientKeyBytesAndProcessBytesHolder is not empty
if(!clientKeyBytesAndProcessBytesHolder.isEmpty()) {
Message message = new Message(partition, clientKeyBytesAndProcessBytesHolder);
// here size of `message.serialize()` byte array should always be less than 50k at all cost
sendToDatabase(message.getAddress(), message.serialize());
}
}
And below is my Message class:
public final class Message {
private final byte dataCenter;
private final byte recordVersion;
private final Map<byte[], byte[]> clientKeyBytesAndProcessBytesHolder;
private final long address;
private final long addressFrom;
private final long addressOrigin;
private final byte recordsPartition;
private final byte replicated;
public Message(Map<byte[], byte[]> clientKeyBytesAndProcessBytesHolder, Partition recordPartition) {
this.clientKeyBytesAndProcessBytesHolder = clientKeyBytesAndProcessBytesHolder;
this.recordsPartition = (byte) recordPartition.getPartition();
this.dataCenter = Utils.CURRENT_LOCATION.get().datacenter();
this.recordVersion = 1;
this.replicated = 0;
long packedAddress = new Data().packAddress();
this.address = packedAddress;
this.addressFrom = 0L;
this.addressOrigin = packedAddress;
}
// Output of this method should always be less than 50k always
public byte[] serialize() {
int bufferCapacity = getBufferCapacity(clientKeyBytesAndProcessBytesHolder); // 36 + dataSize + 1 + 1 + keyLength + 8 + 2;
ByteBuffer byteBuffer = ByteBuffer.allocate(bufferCapacity).order(ByteOrder.BIG_ENDIAN);
// header layout
byteBuffer.put(dataCenter).put(recordVersion).putInt(clientKeyBytesAndProcessBytesHolder.size())
.putInt(bufferCapacity).putLong(address).putLong(addressFrom).putLong(addressOrigin)
.put(recordsPartition).put(replicated);
// now the data layout
for (Map.Entry<byte[], byte[]> entry : clientKeyBytesAndProcessBytesHolder.entrySet()) {
byte keyType = 0;
byte[] key = entry.getKey();
byte[] value = entry.getValue();
byte keyLength = (byte) key.length;
short valueLength = (short) value.length;
ByteBuffer dataBuffer = ByteBuffer.wrap(value);
long timestamp = valueLength > 10 ? dataBuffer.getLong(2) : System.currentTimeMillis();
byteBuffer.put(keyType).put(keyLength).put(key).putLong(timestamp).putShort(valueLength)
.put(value);
}
return byteBuffer.array();
}
private int getBufferCapacity(Map<byte[], byte[]> clientKeyBytesAndProcessBytesHolder) {
int size = 36;
for (Entry<byte[], byte[]> entry : clientKeyBytesAndProcessBytesHolder.entrySet()) {
size += 1 + 1 + 8 + 2;
size += entry.getKey().length;
size += entry.getValue().length;
}
return size;
}
// getters and to string method here
}
Basically, what I have to make sure is whenever the sendToDatabase method is called, size of message.serialize() byte array should always be less than 50k at all cost. My sendToDatabase method sends byte array coming out from serialize method. And because of that condition I am doing below validation plus few other stuff. In the method, I will iterate dataHolders CLQ and I will extract clientKeyBytes and processBytes from it. Here is the validation I am doing:
If the clientKeyBytes length is greater than 255 then I will skip it and continue iterating.
I will keep incrementing the totalSize variable which will be the sum of clientKeyLength and processBytesLength, and this totalSize length should always be less than 50000 bytes.
As soon as it reaches the 50000 limit, I will send the clientKeyBytesAndProcessBytesHolder map to the sendToDatabase method and clear out the map, reset totalSize to 0 and start populating again.
If it doesn't reaches that limit and dataHolders got empty, then it will send whatever it has.
I believe there is some bug in my current code because of which maybe some records are not being sent properly or dropped somewhere because of my condition and I am not able to figure this out. Looks like to properly achieve this 50k condition I may have to use getBufferCapacity method to correctly figure out the size before calling sendToDatabase method?

I checked your code, its look good as per your logic. As you said it will always store the information which is less than 50K but it will actually store information till 50K. To make it less than 50K you have to change the if condition to if (totalSize + additionalLength >= 50000).
If your codes still not fulfilling your requirement i.e. storing information when totalSize + additionalLength is greater than 50k I can advise you few thinks.
As more than 50 threads call this method you need to consider two section in your codes to be synchronize.
One is global variable which is a container dataHoldersByPartition object. If multiple concurrent and parallel searches happened in this container object, outcome might not be perfect. Just check whether container type is synchronized or not. If not make this block like below:-
synchronized(this){
ConcurrentLinkedQueue<DataHolder> dataHolders = dataHoldersByPartition.get(partition);
}
Now, I can give only two suggestion to fix this issue. One is instead of if (totalSize + additionalLength > 50000) this you can check the size of the object clientKeyBytesAndProcessBytesHolder if(sizeof(clientKeyBytesAndProcessBytesHolder) >= 50000) (check appropriate method for sizeof in java). And second one is narrow down the area to check whether it is a side effect of multithreading or not. All these suggestion are to find out the area where exactly problem is and fix should be from your end only.
First check whether you method validateAndSend is exactly satisfying your requirement or not. For that synchronize whole validateAndSend method first and check whether everything fine or still have the same result. If still have the same result that means it is not because of multithreading but your coding is not as per requirement. If its work fine that means it is a problem of multithreading. If method synchronization is fixing your issue but degrade the performance you just remove the synchronization from it and concentrate every small block of your code which might cause the issue and make it synchronize block and remove if still not fixing your issue. Like that finally you locate the block of code which is actually creating the issue and leave it as synchronize to fix it finally.
For example first attempt:-
`private synchronize void validateAndSend`
Second attempts: Remove synchronize key words from the method and do the below step:-
synchronize(this){
Message message = new Message(clientKeyBytesAndProcessBytesHolder, partition);
sendToDatabase(message.getAddress(), message.serialize());
}
If you think that I did not correctly understand you please let me know.

In your validateAndSend I would put whole data to the queue, and do whole processing in separate thread. Please consider command model. That way all threads are going to put their load on queue. Consumer thread has all the data, all the information in place, and can process it quite effectively. The only complicated part is sending response / result back to calling thread. Since in your case that is not a problem - the better. There are some more benefits of this pattern - please look at netflix/hystrix.

Performance difference between assignment and conditional test

This question is specifically geared towards the Java language, but I would not mind feedback about this being a general concept if so. I would like to know which operation might be faster, or if there is no difference between assigning a variable a value and performing tests for values. For this issue we could have a large series of Boolean values that will have many requests for changes. I would like to know if testing for the need to change a value would be considered a waste when weighed against the speed of simply changing the value during every request.
public static void main(String[] args){
Boolean array[] = new Boolean[veryLargeValue];
for(int i = 0; i < array.length; i++) {
array[i] = randomTrueFalseAssignment;
}
for(int i = 400; i < array.length - 400; i++) {
testAndChange(array, i);
}
for(int i = 400; i < array.length - 400; i++) {
justChange(array, i);
}
}
This could be the testAndChange method
public static void testAndChange(Boolean[] pArray, int ind) {
if(pArray)
pArray[ind] = false;
}
This could be the justChange method
public static void justChange(Boolean[] pArray, int ind) {
pArray[ind] = false;
}
If we were to end up with the very rare case that every value within the range supplied to the methods were false, would there be a point where one method would eventually become slower than the other? Is there a best practice for issues similar to this?
Edit: I wanted to add this to help clarify this question a bit more. I realize that the data type can be factored into the answer as larger or more efficient datatypes can be utilized. I am more focused on the task itself. Is the task of a test "if(aConditionalTest)" is slower, faster, or indeterminable without additional informaiton (such as data type) than the task of an assignment "x=avalue".

As #TrippKinetics points out, there is a semantical difference between the two methods. Because you use Boolean instead of boolean, it is possible that one of the values is a null reference. In that case the first method (with the if-statement) will throw an exception while the second, simply assigns values to all the elements in the array.
Assuming you use boolean[] instead of Boolean[]. Optimization is an undecidable problem. There are very rare cases where adding an if-statement could result in better performance. For instance most processors use cache and the if-statement can result in the fact that the executed code is stored exactly on two cache-pages where without an if on more resulting in cache faults. Perhaps you think you will save an assignment instruction but at the cost of a fetch instruction and a conditional instruction (which breaks the CPU pipeline). Assigning has more or less the same cost as fetching a value.
In general however, one can assume that adding an if statement is useless and will nearly always result in slower code. So you can quite safely state that the if statement will slow down your code always.
More specifically on your question, there are faster ways to set a range to false. For instance using bitvectors like:
long[] data = new long[(veryLargeValue+0x3f)>>0x06];//a long has 64 bits
//assign random values
int low = 400>>0x06;
int high = (veryLargeValue-400)>>0x06;
data[low] &= 0xffffffffffffffff<<(0x3f-(400&0x3f));
for(int i = low+0x01; i < high; i++) {
data[i] = 0x00;
}
data[high] &= 0xffffffffffffffff>>(veryLargeValue-400)&0x3f));
The advantage is that a processor can perform operations on 32- or 64-bits at once. Since a boolean is one bit, by storing bits into a long or int, operations are done in parallel.

Java: why is computing faster than assigning value (int)?

The 2 following versions of the same function (which basically tries to recover a password by brute force) do not give same performance:
Version 1:
private static final char[] CHARS = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789".toCharArray();
private static final int N_CHARS = CHARS.length;
private static final int MAX_LENGTH = 8;
private static char[] recoverPassword()
{
char word[];
int refi, i, indexes[];
for (int length = 1; length <= MAX_LENGTH; length++)
{
refi = length - 1;
word = new char[length];
indexes = new int[length];
indexes[length - 1] = 1;
while(true)
{
i = length - 1;
while ((++indexes[i]) == N_CHARS)
{
word[i] = CHARS[indexes[i] = 0];
if (--i < 0)
break;
}
if (i < 0)
break;
word[i] = CHARS[indexes[i]];
if (isValid(word))
return word;
}
}
return null;
}
Version 2:
private static final char[] CHARS = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789".toCharArray();
private static final int N_CHARS = CHARS.length;
private static final int MAX_LENGTH = 8;
private static char[] recoverPassword()
{
char word[];
int refi, i, indexes[];
for (int length = 1; length <= MAX_LENGTH; length++)
{
refi = length - 1;
word = new char[length];
indexes = new int[length];
indexes[length - 1] = 1;
while(true)
{
i = refi;
while ((++indexes[i]) == N_CHARS)
{
word[i] = CHARS[indexes[i] = 0];
if (--i < 0)
break;
}
if (i < 0)
break;
word[i] = CHARS[indexes[i]];
if (isValid(word))
return word;
}
}
return null;
}
I would expect version 2 to be faster, as it does (and that is the only difference):
i = refi;
...as compare to version 1:
i = length -1;
However, it's the opposite: version 1 is faster by over 3%!
Does someone knows why? Is that due to some optimization done by the compiler?
Thank you all for your answers that far.
Just to add that the goal is actually not to optimize this piece of code (which is already quite optimized), but more to understand, from a compiler / CPU / architecture perspective, what could explain such performance difference.
Your answers have been very helpful, thanks again!
Key

It is difficult to check this in a micro-benchmark because you cannot say for sure how the code has been optimised without reading the machine code generated, even then the CPU can do plenty of tricks to optimise it future eg. it turns the x86 code in RISC style instructions to actually execute.
A computation takes as little as one cycle and the CPU can perform up to three of them at once. An access to L1 cache takes 4 cycles and for L2, L3, main memory it takes 11, 40-75, 200 cycles.
Storing values to avoid a simple calculation is actually slower in many cases. BTW using division and modulus is quite expensive and caching this value can be worth it when micro-tuning your code.

The correct answer should be retrievable by a deassembler (i mean .class -> .java converter),
but my guess is that the compiler might have decided to get rid of iref altogether and decided to store length - 1 an auxiliary register.
I'm more of a c++ guy, but I would start by trying:
const int refi = length - 1;
inside the for loop. Also you should probably use
indexes[ refi ] = 1;

Comparing running times of codes does not give exact or quarantine results
First of all, it is not the way comparing performances like this. A running time analysis is needed here. Both 2 codes have same loop structure and their running time are the same. You may have different running times when you run codes. However, they mostly differ with cache hits, I/O times, thread & process schedules. There is no quarantine that code is always completed in a exact time.
However, there is still differences in your code, to understand the difference you should look into your CPU architecture. I can explain it according to x86 architecture basically.
What happens behind the scenes?
i = refi;
CPU takes refi and i to its registers from ram. there is 2 access to ram if the values in not in the cache. and value of i will be written to the ram. However, it always takes different times according to thread & process schedules. Furrhermore, if the values are in virtual memory it wil take longer time.
i = length -1;
CPU also access i and length from ram or cache. there is same number of accesses. In addition, there is a subtraction here which means extra CPU cycles. That is why you think this one take longer time to complete. It is expected, but the issues that i mentioned above explain why this take longer time.
Summation
As i explain this is not the way of comparing performance. I think, there is no real difference between these codes. There are lots of optimizations inside CPU and also in compiler. You can see optimized codes if you decompile .class files.
My advice is it is better to minimize BigO running time analysis. If you find better algorithms it is the best way of optimizing codes. In case you still have bottlenecks in your code, you may try micro-benchmarking.
See also
Analysis of algorithms
Big O notation
Microprocessor
Compiler optimization
CPU Scheduling

To start with, you can't really compare the performance by just running your program - micro benchmarking in Java is complicated.
Also, a subtraction on modern CPUs can take as little as a third of a clock cycle on average. On a 3GHz CPU, that is 0.1 nanoseconds. And nothing tells you that the subtraction actually happens as the compiler might have modified the code.
So:
You should try to check the generated assembly code.
If you really care about the performance, create an appropriate micro-benchmark.

Why is System.out.println so slow?

Is this something common to all programming languages? Doing multiple print followed by a println seems faster but moving everything to a string and just printing that seems fastest. Why?
EDIT: For example, Java can find all the prime numbers up to 1 million in less than a second - but printing then all out each on their own println can take minutes! Up to a 10 billion can hours to print!
EX:
package sieveoferatosthenes;
public class Main {
public static void main(String[] args) {
int upTo = 10000000;
boolean primes[] = new boolean[upTo];
for( int b = 0; b < upTo; b++ ){
primes[b] = true;
}
primes[0] = false;
primes[1] = false;
int testing = 1;
while( testing <= Math.sqrt(upTo)){
testing ++;
int testingWith = testing;
if( primes[testing] ){
while( testingWith < upTo ){
testingWith = testingWith + testing;
if ( testingWith >= upTo){
}
else{
primes[testingWith] = false;
}
}
}
}
for( int b = 2; b < upTo; b++){
if( primes[b] ){
System.out.println( b );
}
}
}
}

println is not slow, it's the underlying PrintStream that is connected with the console, provided by the hosting operating system.
You can check it yourself: compare dumping a large text file to the console with piping the same textfile into another file:
cat largeTextFile.txt
cat largeTextFile.txt > temp.txt
Reading and writing are similiar and proportional to the size of the file (O(n)), the only difference is, that the destination is different (console compared to file). And that's basically the same with System.out.
The underlying OS operation (displaying chars on a console window) is slow because
The bytes have to be sent to the console application (should be quite fast)
Each char has to be rendered using (usually) a true type font (that's pretty slow, switching off anti aliasing could improve performance, btw)
The displayed area may have to be scrolled in order to append a new line to the visible area (best case: bit block transfer operation, worst case: re-rendering of the complete text area)

System.out is a static PrintStream class. PrintStream has, among other things, those methods you're probably quite familiar with, like print() and println() and such.
It's not unique to Java that input and output operations take a long time. "long." printing or writing to a PrintStream takes a fraction of a second, but over 10 billion instances of this print can add up to quite a lot!
This is why your "moving everything to a String" is the fastest. Your huge String is built, but you only print it once. Sure, it's a huge print, but you spend time on actually printing, not on the overhead associated with the print() or println().
As Dvd Prd has mentioned, Strings are immutable. That means whenever you assign a new String to an old one but reusing references, you actually destroy the reference to the old String and create a reference to the new one. So you can make this whole operation go even faster by using the StringBuilder class, which is mutable. This will decrease the overhead associated with building that string you'll eventually print.

I believe this is because of buffering. A quote from the article:
Another aspect of buffering concerns
text output to a terminal window. By
default, System.out (a PrintStream) is
line buffered, meaning that the output
buffer is flushed when a newline
character is encountered. This is
important for interactivity, where
you'd like to have an input prompt
displayed before actually entering any
input.
A quote explaining buffers from wikipedia:
In computer science, a buffer is a
region of memory used to temporarily
hold data while it is being moved from
one place to another. Typically, the
data is stored in a buffer as it is
retrieved from an input device (such
as a Mouse) or just before it is sent
to an output device (such as Speakers)
public void println()
Terminate the current line by writing
the line separator string. The line
separator string is defined by the
system property line.separator, and is
not necessarily a single newline
character ('\n').
So the buffer get's flushed when you do println which means new memory has to be allocated etc which makes printing slower. The other methods you specified require lesser flushing of buffers thus are faster.

Take a look at my System.out.println replacement.
By default, System.out.print() is only line-buffered and does a lot work related to Unicode handling. Because of its small buffer size, System.out.println() is not well suited to handle many repetitive outputs in a batch mode. Each line is flushed right away. If your output is mainly ASCII-based then by removing the Unicode-related activities, the overall execution time will be better.

If you're printing to the console window, not to a file, that will be the killer.
Every character has to be painted, and on every line the whole window has to be scrolled.
If the window is partly overlaid with other windows, it also has to do clipping.
That's going to take far more cycles than what your program is doing.
Usually that's not a bad price to pay, since console output is supposed to be for your reading pleasure :)

The problem you have is that displaying to the screen is very espensive, especially if you have a graphical windows/X-windows environment (rather than a pure text terminal) Just to render one digit in a font is far more expensive than the calculations you are doing. When you send data to the screen faster than it can display it, it buffered the data and quickly blocks. Even writing to a file is significant compare to the calculations, but its 10x - 100x faster than displaying on the screen.
BTW: math.sqrt() is very expensive, and using a loop is much slower than using modulus i.e. % to determine if a number is a multiple. BitSet can be 8x more space efficient than boolean[], and faster for operations on multiple bits e.g. counting or searching bits.
If I dump the output to a file, it is quick, but writing to the console is slow, and if I write to the console the data which was written to a file it takes about the same amount of time.
Took 289 ms to examine 10,000,000 numbers.
Took 149 ms to toString primes up to 10,000,000.
Took 306 ms to write to a file primes up to 10,000,000.
Took 61,082 ms to write to a System.out primes up to 10,000,000.
time cat primes.txt
real 1m24.916s
user 0m3.619s
sys 0m12.058s
The code
int upTo = 10*1000*1000;
long start = System.nanoTime();
BitSet nonprimes = new BitSet(upTo);
for (int t = 2; t * t < upTo; t++) {
if (nonprimes.get(t)) continue;
for (int i = 2 * t; i <= upTo; i += t)
nonprimes.set(i);
}
PrintWriter report = new PrintWriter("report.txt");
long time = System.nanoTime() - start;
report.printf("Took %,d ms to examine %,d numbers.%n", time / 1000 / 1000, upTo);
long start2 = System.nanoTime();
for (int i = 2; i < upTo; i++) {
if (!nonprimes.get(i))
Integer.toString(i);
}
long time2 = System.nanoTime() - start2;
report.printf("Took %,d ms to toString primes up to %,d.%n", time2 / 1000 / 1000, upTo);
long start3 = System.nanoTime();
PrintWriter pw = new PrintWriter(new BufferedOutputStream(new FileOutputStream("primes.txt"), 64*1024));
for (int i = 2; i < upTo; i++) {
if (!nonprimes.get(i))
pw.println(i);
}
pw.close();
long time3 = System.nanoTime() - start3;
report.printf("Took %,d ms to write to a file primes up to %,d.%n", time3 / 1000 / 1000, upTo);
long start4 = System.nanoTime();
for (int i = 2; i < upTo; i++) {
if (!nonprimes.get(i))
System.out.println(i);
}
long time4 = System.nanoTime() - start4;
report.printf("Took %,d ms to write to a System.out primes up to %,d.%n", time4 / 1000 / 1000, upTo);
report.close();

Most of the answers here are right, but they don't cover the most important point: system calls. This is the operation that induces the more overhead.
When your software needs to access some hardware resource (your screen for example), it needs to ask the OS (or hypervisor) if it can access the hardware. This costs a lot:
Here are interesting blogs about syscalls, the last one being dedicated to syscall and Java
http://arkanis.de/weblog/2017-01-05-measurements-of-system-call-performance-and-overhead
http://www.brendangregg.com/blog/2014-05-11/strace-wow-much-syscall.html
https://blog.packagecloud.io/eng/2017/03/14/using-strace-to-understand-java-performance-improvement/

Efficient way to search a stream for a string

Let's suppose that have a stream of text (or Reader in Java) that I'd like to check for a particular string. The stream of text might be very large so as soon as the search string is found I'd like to return true and also try to avoid storing the entire input in memory.
Naively, I might try to do something like this (in Java):
public boolean streamContainsString(Reader reader, String searchString) throws IOException {
char[] buffer = new char[1024];
int numCharsRead;
while((numCharsRead = reader.read(buffer)) > 0) {
if ((new String(buffer, 0, numCharsRead)).indexOf(searchString) >= 0)
return true;
}
return false;
}
Of course this fails to detect the given search string if it occurs on the boundary of the 1k buffer:
Search text: "stackoverflow"
Stream buffer 1: "abc.........stack"
Stream buffer 2: "overflow.......xyz"
How can I modify this code so that it correctly finds the given search string across the boundary of the buffer but without loading the entire stream into memory?
Edit: Note when searching a stream for a string, we're trying to minimise the number of reads from the stream (to avoid latency in a network/disk) and to keep memory usage constant regardless of the amount of data in the stream. Actual efficiency of the string matching algorithm is secondary but obviously, it would be nice to find a solution that used one of the more efficient of those algorithms.

There are three good solutions here:
If you want something that is easy and reasonably fast, go with no buffer, and instead implement a simple nondeterminstic finite-state machine. Your state will be a list of indices into the string you are searching, and your logic looks something like this (pseudocode):
String needle;
n = needle.length();
for every input character c do
add index 0 to the list
for every index i in the list do
if c == needle[i] then
if i + 1 == n then
return true
else
replace i in the list with i + 1
end
else
remove i from the list
end
end
end
This will find the string if it exists and you will never need a
buffer.
Slightly more work but also faster: do an NFA-to-DFA conversion that figures out in advance what lists of indices are possible, and assign each one to a small integer. (If you read about string search on Wikipedia, this is called the powerset construction.) Then you have a single state and you make a state-to-state transition on each incoming character. The NFA you want is just the DFA for the string preceded with a state that nondeterministically either drops a character or tries to consume the current character. You'll want an explicit error state as well.
If you want something faster, create a buffer whose size is at least twice n, and user Boyer-Moore to compile a state machine from needle. You'll have a lot of extra hassle because Boyer-Moore is not trivial to implement (although you'll find code online) and because you'll have to arrange to slide the string through the buffer. You'll have to build or find a circular buffer that can 'slide' without copying; otherwise you're likely to give back any performance gains you might get from Boyer-Moore.

I did a few changes to the Knuth Morris Pratt algorithm for partial searches. Since the actual comparison position is always less or equal than the next one there is no need for extra memory. The code with a Makefile is also available on github and it is written in Haxe to target multiple programming languages at once, including Java.
I also wrote a related article: searching for substrings in streams: a slight modification of the Knuth-Morris-Pratt algorithm in Haxe. The article mentions the Jakarta RegExp, now retired and resting in the Apache Attic. The Jakarta Regexp library “match” method in the RE class uses a CharacterIterator as a parameter.
class StreamOrientedKnuthMorrisPratt {
var m: Int;
var i: Int;
var ss:
var table: Array<Int>;
public function new(ss: String) {
this.ss = ss;
this.buildTable(this.ss);
}
public function begin() : Void {
this.m = 0;
this.i = 0;
}
public function partialSearch(s: String) : Int {
var offset = this.m + this.i;
while(this.m + this.i - offset < s.length) {
if(this.ss.substr(this.i, 1) == s.substr(this.m + this.i - offset,1)) {
if(this.i == this.ss.length - 1) {
return this.m;
}
this.i += 1;
} else {
this.m += this.i - this.table[this.i];
if(this.table[this.i] > -1)
this.i = this.table[this.i];
else
this.i = 0;
}
}
return -1;
}
private function buildTable(ss: String) : Void {
var pos = 2;
var cnd = 0;
this.table = new Array<Int>();
if(ss.length > 2)
this.table.insert(ss.length, 0);
else
this.table.insert(2, 0);
this.table[0] = -1;
this.table[1] = 0;
while(pos < ss.length) {
if(ss.substr(pos-1,1) == ss.substr(cnd, 1))
{
cnd += 1;
this.table[pos] = cnd;
pos += 1;
} else if(cnd > 0) {
cnd = this.table[cnd];
} else {
this.table[pos] = 0;
pos += 1;
}
}
}
public static function main() {
var KMP = new StreamOrientedKnuthMorrisPratt("aa");
KMP.begin();
trace(KMP.partialSearch("ccaabb"));
KMP.begin();
trace(KMP.partialSearch("ccarbb"));
trace(KMP.partialSearch("fgaabb"));
}
}

The Knuth-Morris-Pratt search algorithm never backs up; this is just the property you want for your stream search. I've used it before for this problem, though there may be easier ways using available Java libraries. (When this came up for me I was working in C in the 90s.)
KMP in essence is a fast way to build a string-matching DFA, like Norman Ramsey's suggestion #2.

This answer applied to the initial version of the question where the key was to read the stream only as far as necessary to match on a String, if that String was present. This solution would not meet the requirement to guarantee fixed memory utilisation, but may be worth considering if you have found this question and are not bound by that constraint.
If you are bound by the constant memory usage constraint, Java stores arrays of any type on the heap, and as such nulling the reference does not deallocate memory in any way; I think any solution involving arrays in a loop will consume memory on the heap and require GC.
For simple implementation, maybe Java 5's Scanner which can accept an InputStream and use a java.util.regex.Pattern to search the input for might save you worrying about the implementation details.
Here's an example of a potential implementation:
public boolean streamContainsString(Reader reader, String searchString)
throws IOException {
Scanner streamScanner = new Scanner(reader);
if (streamScanner.findWithinHorizon(searchString, 0) != null) {
return true;
} else {
return false;
}
}
I'm thinking regex because it sounds like a job for a Finite State Automaton, something that starts in an initial state, changing state character by character until it either rejects the string (no match) or gets to an accept state.
I think this is probably the most efficient matching logic you could use, and how you organize the reading of the information can be divorced from the matching logic for performance tuning.
It's also how regexes work.

Instead of having your buffer be an array, use an abstraction that implements a circular buffer. Your index calculation will be buf[(next+i) % sizeof(buf)], and you'll have to be careful to full the buffer one-half at a time. But as long as the search string fits in half the buffer, you'll find it.

I believe the best solution to this problem is to try to keep it simple. Remember, beacause I'm reading from a stream, I want to keep the number of reads from the stream to a minimum (as network or disk latency may be an issue) while keeping the amount of memory used constant (as the stream may be very large in size). Actual efficiency of the string matching is not the number one goal (as that has been studied to death already).
Based on AlbertoPL's suggestion, here's a simple solution that compares the buffer against the search string character by character. The key is that because the search is only done one character at a time, no back tracking is needed and therefore no circular buffers, or buffers of a particular size are needed.
Now, if someone can come up with a similar implementation based on Knuth-Morris-Pratt search algorithm then we'd have a nice efficient solution ;)
public boolean streamContainsString(Reader reader, String searchString) throws IOException {
char[] buffer = new char[1024];
int numCharsRead;
int count = 0;
while((numCharsRead = reader.read(buffer)) > 0) {
for (int c = 0; c < numCharsRead; c++) {
if (buffer[c] == searchString.charAt(count))
count++;
else
count = 0;
if (count == searchString.length()) return true;
}
}
return false;
}

If you're not tied to using a Reader, then you can use Java's NIO API to efficiently load the file. For example (untested, but should be close to working):
public boolean streamContainsString(File input, String searchString) throws IOException {
Pattern pattern = Pattern.compile(Pattern.quote(searchString));
FileInputStream fis = new FileInputStream(input);
FileChannel fc = fis.getChannel();
int sz = (int) fc.size();
MappedByteBuffer bb = fc.map(FileChannel.MapMode.READ_ONLY, 0, sz);
CharsetDecoder decoder = Charset.forName("UTF-8").newDecoder();
CharBuffer cb = decoder.decode(bb);
Matcher matcher = pattern.matcher(cb);
return matcher.matches();
}
This basically mmap()'s the file to search and relies on the operating system to do the right thing regarding cache and memory usage. Note however that map() is more expensive the just reading the file in to a large buffer for files less than around 10 KiB.

A very fast searching of a stream is implemented in the RingBuffer class from the Ujorm framework. See the sample:
Reader reader = RingBuffer.createReader("xxx ${abc} ${def} zzz");
String word1 = RingBuffer.findWord(reader, "${", "}");
assertEquals("abc", word1);
String word2 = RingBuffer.findWord(reader, "${", "}");
assertEquals("def", word2);
String word3 = RingBuffer.findWord(reader, "${", "}");
assertEquals("", word3);
The single class implementation is available on the SourceForge:
For more information see the link.

Implement a sliding window. Have your buffer around, move all elements in the buffer one forward and enter a single new character in the buffer at the end. If the buffer is equal to your searched word, it is contained.
Of course, if you want to make this more efficient, you can look at a way to prevent moving all elements in the buffer around, for example by having a cyclic buffer and a representation of the strings which 'cycles' the same way the buffer does, so you only need to check for content-equality. This saves moving all elements in the buffer.

I think you need to buffer a small amount at the boundary between buffers.
For example if your buffer size is 1024 and the length of the SearchString is 10, then as well as searching each 1024-byte buffer you also need to search each 18-byte transition between two buffers (9 bytes from the end of the previous buffer concatenated with 9 bytes from the start of the next buffer).

I'd say switch to a character by character solution, in which case you'd scan for the first character in your target text, then when you find that character increment a counter and look for the next character. Every time you don't find the next consecutive character restart the counter. It would work like this:
public boolean streamContainsString(Reader reader, String searchString) throws IOException {
char[] buffer = new char[1024];
int numCharsRead;
int count = 0;
while((numCharsRead = reader.read(buffer)) > 0) {
if (buffer[numCharsRead -1] == searchString.charAt(count))
count++;
else
count = 0;
if (count == searchString.size())
return true;
}
return false;
}
The only problem is when you're in the middle of looking through characters... in which case there needs to be a way of remembering your count variable. I don't see an easy way of doing so except as a private variable for the whole class. In which case you would not instantiate count inside this method.

You might be able to implement a very fast solution using Fast Fourier Transforms, which, if implemented properly, allow you to do string matching in times O(nlog(m)), where n is the length of the longer string to be matched, and m is the length of the shorter string. You could, for example, perform FFT as soon as you receive an stream input of length m, and if it matches, you can return, and if it doesn't match, you can throw away the first character in the stream input, wait for a new character to appear through the stream, and then perform FFT again.

You can increase the speed of search for very large strings by using some string search algorithm

If you're looking for a constant substring rather than a regex, I'd recommend Boyer-Moore. There's plenty of source code on the internet.
Also, use a circular buffer, to avoid think too hard about buffer boundaries.
Mike.

I also had a similar problem: skip bytes from the InputStream until specified string (or byte array). This is the simple code based on circular buffer. It is not very efficient but works for my needs:
private static boolean matches(int[] buffer, int offset, byte[] search) {
final int len = buffer.length;
for (int i = 0; i < len; ++i) {
if (search[i] != buffer[(offset + i) % len]) {
return false;
}
}
return true;
}
public static void skipBytes(InputStream stream, byte[] search) throws IOException {
final int[] buffer = new int[search.length];
for (int i = 0; i < search.length; ++i) {
buffer[i] = stream.read();
}
int offset = 0;
while (true) {
if (matches(buffer, offset, search)) {
break;
}
buffer[offset] = stream.read();
offset = (offset + 1) % buffer.length;
}
}

Here is my implementation:
static boolean containsKeywordInStream( Reader ir, String keyword, int bufferSize ) throws IOException{
SlidingContainsBuffer sb = new SlidingContainsBuffer( keyword );
char[] buffer = new char[ bufferSize ];
int read;
while( ( read = ir.read( buffer ) ) != -1 ){
if( sb.checkIfContains( buffer, read ) ){
return true;
}
}
return false;
}
SlidingContainsBuffer class:
class SlidingContainsBuffer{
private final char[] keyword;
private int keywordIndexToCheck = 0;
private boolean keywordFound = false;
SlidingContainsBuffer( String keyword ){
this.keyword = keyword.toCharArray();
}
boolean checkIfContains( char[] buffer, int read ){
for( int i = 0; i < read; i++ ){
if( keywordFound == false ){
if( keyword[ keywordIndexToCheck ] == buffer[ i ] ){
keywordIndexToCheck++;
if( keywordIndexToCheck == keyword.length ){
keywordFound = true;
}
} else {
keywordIndexToCheck = 0;
}
} else {
break;
}
}
return keywordFound;
}
}
This answer fully qualifies the task:
The implementation is able to find the searched keyword even if it was split between buffers
Minimum memory usage defined by the buffer size
Number of reads will be minimized by using bigger buffer