Persistent connections versus pooling - php framework
When the traffic to one or more php application is high, the application may open several connections to the database server using tge same credentials. The database server may reject these saying that there are too many connections open and the application throws an exception saying CDBconnection could not open a connection. In such cases, the database server expects the application to not open new connections but to use some form of connection pooling. Pooling helps conserve the resources which would otherwise have been wasted one for every connection. Also, it makes the application run fast as the coonections dont have to be opened on every request. In addition to max_connections parameter to pooling, another parameter that plays important role is thread_cache_size. Requests for threads are satisfied by reusing threads taken from the cache whenever possible and only if the cache is enpty, a new thread is created. Typically, this variable is set so that tge fraction of requests for wgich threads are newly created is rather small.
Consequently in a high traffic server this maybe setvto a arbitrarily high value
However,
Php framework operates on a request basis. It does not support connection pooling. The alternatives are to use something else that the framework provides as we cannot use anything outside the framework. One such technique is to use persisted connections. We set this property to true in the application config file under the db section.
A caveat with this approach is that it does not address any issues originating from the sql used with a connection by the user or application. If there's a bug in the sql wjich leads to long running queries, then this won't mitigate that. However, the first step is to lower the number of connections opened before analysing any locks held and orphaned by SQL. It is therefore prudent to use register_shutdown_function that cleansup after each SQL execution.
# a microservice example
https://drive.google.com/file/d/0B-H0-bJJQAq5eFhuYjIyQXNFR3c/view?usp=sharing
https://drive.google.com/file/d/0B-H0-bJJQAq5aFlQLS0wZ3pPc28/view?usp=sharing
#codingexercise
Find if there exists a pair in an array that can add up to a sum that us divisible by k
We gave a few solutions yesterday.
Another approach for small range of numbers in the list is as follows:
bool hasPairDivisibleByK(List<int> nums, int k)
{
Assert (nums.all(x => x > 0));
Assert( k > 0);
Var min = nums.min();
Var max = nums.max();
Int low = (min + min ) /k;
Int high = (max + max) /k;
nums.sort();
For (int I = low; I <= high+1; i++)
If (hasPairSum(nums, I * k) == true)
Return true;
Return false;
}
bool hasPairSum(List<int> nums, int sum)
{
int left = 0;
int right = nums.Count -1;
if (nums.Count < 2 ) return false;
// nums.Sort();
while (left < right)
{
if(nums[left] + nums[right] == sum)
return true;
else if (nums[left] + nums[right] < sum)
left++;
else
right--;
}
return false;
}
Find minimum distance between two numbers in an array
Int getMinDistance(list<int> nums, int X, int Y)
{
Int min = int_max;
For(int i = 0; i < nums.count; i++)
For(int j = i + 1; i < nums.count; i++)
When the traffic to one or more php application is high, the application may open several connections to the database server using tge same credentials. The database server may reject these saying that there are too many connections open and the application throws an exception saying CDBconnection could not open a connection. In such cases, the database server expects the application to not open new connections but to use some form of connection pooling. Pooling helps conserve the resources which would otherwise have been wasted one for every connection. Also, it makes the application run fast as the coonections dont have to be opened on every request. In addition to max_connections parameter to pooling, another parameter that plays important role is thread_cache_size. Requests for threads are satisfied by reusing threads taken from the cache whenever possible and only if the cache is enpty, a new thread is created. Typically, this variable is set so that tge fraction of requests for wgich threads are newly created is rather small.
Consequently in a high traffic server this maybe setvto a arbitrarily high value
However,
Php framework operates on a request basis. It does not support connection pooling. The alternatives are to use something else that the framework provides as we cannot use anything outside the framework. One such technique is to use persisted connections. We set this property to true in the application config file under the db section.
A caveat with this approach is that it does not address any issues originating from the sql used with a connection by the user or application. If there's a bug in the sql wjich leads to long running queries, then this won't mitigate that. However, the first step is to lower the number of connections opened before analysing any locks held and orphaned by SQL. It is therefore prudent to use register_shutdown_function that cleansup after each SQL execution.
# a microservice example
https://drive.google.com/file/d/0B-H0-bJJQAq5eFhuYjIyQXNFR3c/view?usp=sharing
https://drive.google.com/file/d/0B-H0-bJJQAq5aFlQLS0wZ3pPc28/view?usp=sharing
#codingexercise
Find if there exists a pair in an array that can add up to a sum that us divisible by k
We gave a few solutions yesterday.
Another approach for small range of numbers in the list is as follows:
bool hasPairDivisibleByK(List<int> nums, int k)
{
Assert (nums.all(x => x > 0));
Assert( k > 0);
Var min = nums.min();
Var max = nums.max();
Int low = (min + min ) /k;
Int high = (max + max) /k;
nums.sort();
For (int I = low; I <= high+1; i++)
If (hasPairSum(nums, I * k) == true)
Return true;
Return false;
}
bool hasPairSum(List<int> nums, int sum)
{
int left = 0;
int right = nums.Count -1;
if (nums.Count < 2 ) return false;
// nums.Sort();
while (left < right)
{
if(nums[left] + nums[right] == sum)
return true;
else if (nums[left] + nums[right] < sum)
left++;
else
right--;
}
return false;
}
Find minimum distance between two numbers in an array
Int getMinDistance(list<int> nums, int X, int Y)
{
Int min = int_max;
For(int i = 0; i < nums.count; i++)
For(int j = i + 1; i < nums.count; i++)
{
If (((nums[i] == X && nums[j] == Y) ||
(nums[i] == X && nums[j] == Y)) &&
Math.abs(i-j) < min)
Min = math.abs(i-j);
}
Return min;