#include <bits/stdc++.h>
using namespace std;
void solve() {
int N;
cin >> N;
vector<pair<long long,long long>> c(N + 1);
vector<long long> pref(N + 1);
for (int i = 1; i <= N; i++) {
cin >> c[i].first >> c[i].second;
}
sort(c.begin() + 1, c.end());
for (int i = 1; i <= N; i++) {
pref[i] = pref[i - 1] + c[i].second;
}
long long mx = -2E18, ans = 0;
for (int i = N; i >= 1; i--) {
ans = max(ans, mx - (pref[i] - c[i].first - c[i].second));
mx = max(mx, pref[i] - c[i].first);
}
cout << ans << '\n';
}
int main() {
ios_base::sync_with_stdio(false);
cin.tie(nullptr);
int tests = 1;
// cin >> tests;
while (tests--) solve();
return 0;
}
/*
sort by increasing sizes
the optimal answer will be a subarray ... (N ^ 2 is immediate)
oh!
pref[r] - pref[l - 1] - (A[r] - A[l])
(pref[r] - A[r]) - (pref[l - 1] - A[l])
(pref[r] - A[r]) - (pref[l] - A[l] - B[l])
(pref[r] - A[r]) - (pref[l] - A[l]) + B[l]
ok now this should be easy.
*/
