Eficiency and Low Cost: Pro Tips for you to save 50% of your money with Google Cloud

1. P R O T I P S F O R Y O U T O S AV E 5 0 % O F
Y O U R M O N E Y W I T H G O O G L E C L O U D
E F F I C I E N C Y A N D L O W C O S T
D A N I E L C U K I E R ( G D E )
@ D A N I C U K I

3. http://www.flickr.com/photos/zerial/2804865447/sizes/z/in/photostream/

4. http://www.flickr.com/photos/bike/266453254/sizes/l/in/photostream/

5. http://www.flickr.com/photos/tim_d/184018928/sizes/l/in/photostream/

6. http://www.flickr.com/photos/stars6/4381851322/sizes/l/in/photostream/

7. W H AT I F…

8. Click!

10. (╯°□°)╯︵ ┻━┻

11. Data Center
Colocation
Physical Servers
Virtual Servers
Managed Service
Containers
Managed Container
App Engine Flex
App Engine Standard
SaaS
Praia
Low cost, high work
High cost, low work
W H E R E A R E Y O U ?

12. P R O T I P S F O R Y O U T O S AV E 5 0 % O F
Y O U R M O N E Y W I T H G O O G L E C L O U D
E F I C I E N C Y A N D L O W C O S T
D A N I E L C U K I E R ( G D E )
@ D A N I C U K I

13. S U S TA I N E D U S A G E

14. C O M M I T E D U S E
MACHINE
TYPE
ITEM FULL PRICE
1 YEAR
COMMITMENT
3 YEAR
COMMITMENT
CUSTOM
MACHINE
TYPES
VCPU $0.033174 /
VCPU HOUR
$0.019915 /
VCPU HOUR
$0.014225 /
VCPU HOUR
MEMORY $0.004446 / GB
HOUR
$0.002669 / GB
HOUR
$0.001907 / GB
HOUR
PREDEFINED
MACHINE
TYPES*
N1-
STANDARD-16
$0.7600 / HOUR $0.47878 / HOUR $0.34202 / HOUR
N1-
HIGHMEM-16
$0.9472 / HOUR $0.59622 / HOUR $0.42593 / HOUR
N1-
HIGHCPU-16
$0.5672 / HOUR $0.35707 / HOUR $0.25506 / HOUR

15. P R E E M P T I B L E
http://www.agileandart.com/2016/05/04/how-i-reduced-my-cloud-cost-using-google-cloud/

16. S T E P 1 : S E T U P T H E C L I E N T
T O B E FA U LT T O L E R A N T
def query(params, retries = 0) {
val response = api.query(params)
response.onSuccess {
codeForSuccess()
}
response.onFailure {
case x => {
LOG.error(s"Failure on $retries try of API request: " + x.getMessage)
Thread.sleep(retries * 3000) // this sleep is optional
query(params, retries + 1) // could have a maximum
// number of retries here
}
}
}
P R E E M P T I B L E

17. S T E P 2 : P U T A L L S E R V E R S
B E H I N D A L O A D B A L A N C E R
#!/bin/bash
SERVER_SUBSTRING=playax-fingerprint
EMPTY_FILE=`cat /etc/haproxy/haproxy.cfg |grep -v $SERVER_SUBSTRING`
NEW_LINES=`gcloud compute instances list |grep $SERVER_SUBSTRING | sed 's/
true//g' |sed 's/ [ ]*/ /g'|cut -d" " -f4|awk '{print " server playax-
fingerprint" $NF " " $NF ":9000 check inter 5s rise 1 fall 1 weight 1"}'`
echo "$EMPTY_FILE" >new_config
echo "$NEW_LINES" >>new_config
sudo cp new_config /etc/haproxy/haproxy.cfg
sudo ./restart.sh
P R E E M P T I B L E

18. S T E P 3 : C R E AT E A N I N S TA N C E
G R O U P F O R T H E S E S E R V E R S
P R E E M P T I B L E

19. P R E E M P T I B L E

20. P R E E M P T I B L E

21. P R E E M P T I B L E
Price reduction: $490 or 48.6%
BEFORE AFTER
SERVERS 20 24
COST PER SERVER $0.07 $0.03
TOTAL COST PER HOUR $1.4 $0.72
TOTAL COST PER MONTH $1,008 $518

22. F I L E S T O R A G E

23. Public Processed Long Term Storage
Characteristics 99.95% availability
Geo-redundant
99.9% availability
Data stored in a
narrow
geographic region
99.0% availability
30-day minimum
storage duration
99.0% availability
90-day minimum
storage duration
Uses Cases Frequently
accessed
World Audience
Website content
Streaming video
Gaming
Mobile
Accessed by your
system
But not by your
customers
Big Data sets for
processing
Backup
Long tail
multimedia
Disaster recovery
Compliance
archives
Storage Cost
GB/month
$.026 $.02 $.01 $.007
Retrieval Cost
GB
$.01 $.05
Storage Type Multi-region Regional Nearline Coldline
F I L E S T O R A G E

24. E X A M P L E
$ g s u t i l d u - s g s : / / p l a y a x - re d u c e d / p ro d u c t i o n / d e t e c t i o n s
F I L E S T O R A G E
1 4 , 7 1 9 , 8 0 0 , 9 6 2 , 9 0 4 = 1 3 , 7 0 8 G B
Storage Type Multi-region Regional Nearline Coldline
Cost GB/month $0.026 $0.02 $0.01 $0.007
Cost 2 years of data / month $356 $274 $137 $95
Cost 3 months of data / month $44 $34 $17 $12

25. P R E E M P T I B L E
Price reduction: $339 or 95.2%
BEFORE AFTER
GB OF DATA 13,708 2
COST / GB $0.026 $1000
COST / MONTH $356 $17

26. C O D E M AT T E R S

27. 2 9 6 7 5 , 1 4 3 7 6 , 5 5 9 5 0 , 2 5 8 6 , 8 0 9 9 5 , 3 7 7 5 9 , 6 5 8 4 5 , 3 1 5 6 2 ,
1 5 8 0 0 , 4 2 7 7 5 , 9 6 5 4 2 , 2 1 4 8 5 , 9 3 7 2 8 , 5 9 0 2 2 , 7 8 8 5 4 , 9 6 6 8 ,
3 1 0 8 8 , 4 4 0 0 7 , 4 1 8 7 8 , 2 7 6 8 0 , 5 4 4 5 0 , 9 1 3 3 6 , 1 9 2 8 1 , 1 7 0 2 5 ,
7 4 9 7 8 , 3 3 9 4 2 , 2 4 6 8 9 , 2 7 2 9 1 , 5 7 0 2 4 , 3 1 9 4 7 , 9 5 7 3 9 , 9 0 7 9 4 ,
2 8 6 2 4 , 7 6 6 0 2 , 5 9 6 2 7 , 4 8 0 7 7 , 6 4 4 6 4 , 2 7 6 2 8 , 2 2 9 9 , 4 0 9 9 8 ,
3 6 0 9 9 , 1 7 9 1 , 2 1 5 6 3 , 2 9 0 7 8 , 5 8 8 4 4 , 3 2 9 5 1 , 8 1 1 0 5 , 9 9 0 3 1 ,
2 2 3 8 9 , 2 2 5 4 9 , 2 8 6 9 , 2 4 2 5 3 , 6 8 3 1 8 , 5 7 6 7 8 , 5 8 6 0 5 , 8 5 8 3 4 ,
7 9 2 3 2 , 7 6 8 9 5 , 8 4 7 1 9 , 9 1 1 1 , 9 4 2 8 5 , 8 2 6 3 9 , 3 2 2 2 6 , 2 7 2 2 4 ,
1 6 7 4 0 , 4 2 4 6 8 , 4 7 6 8 8 , 4 2 3 9 4 , 9 3 1 5 0 , 5 3 5 1 0 , 7 9 2 3 0 , 4 3 8 6 0 ,
4 2 1 0 5 , 7 3 3 8 3 , 6 4 7 1 0 , 7 2 6 6 , 3 9 3 1 3 , 8 2 2 3 5 , 9 7 6 4 9 , 4 6 6 2 6 ,
8 3 9 2 3 , 7 7 3 9 4 , 8 6 8 1 7 , 9 3 5 0 1 , 3 5 2 9 6 , 5 7 0 6 9 , 6 8 8 2 3 , 7 5 3 6 7 ,
6 0 1 6 5 , 5 9 4 9 9 , 5 5 5 6 1 , 7 1 7 8 7 , 7 3 8 4 5 , 9 8 8 8 , 3 6 2 5 7 , 2 0 8 4 4 ,
8 9 0 4 9 , 6 7 7 2 0 , 3 6 3 2 1 , 6 5 2 9 9 , 1 0 1 2 1 , 4 0 7 8 3 , 5 7 6 5 , 9 6 2 9 2 ,
5 0 6 5 , 1 2 1 8 7 , 8 2 2 3 5 , 3 2 9 3 1 , 3 7 4 0 4 , 6 5 9 5 6 , 8 0 9 8 3 , 8 9 4 3 7 ,
8 7 5 3 , 6 2 9 1 0 , 2 9 6 5 2 , 9 2 5 2 4 , 7 8 7 3 6 , 1 6 9 6 1 , 4 7 4 9 6 , 8 9 2 7 7 ,
6 6 1 0 2 , 5 6 4 9 2 , 3 2 0 4 2 , 6 1 3 5 6 , 9 7 8 5 3 , 8 0 2 7 8 , 8 7 5 5 9 , 8 2 0 2 4 ,
6 6 5 5 0 , 8 1 5 0 0 , 5 7 9 5 9 , 6 4 8 7 1 , 7 0 4 7 0 , 2 4 8 8 6 , 4 2 6 2 6 , 9 3 5 9 8 ,
7 1 6 3 , 6 9 7 7 5 , 4 4 9 1 5 , 9 7 5 3 1 , 6 1 4 3 , 9 1 5 1 , 6 7 0 3 , 4 2 1 5 6 , 2 8 6 3 6 ,
8 1 2 4 9 , 3 0 4 6 6 , 6 8 5 0 6 , 9 8 9 7 0 , 6 3 8 0 9 , 6 6 0 0 1 , 4 8 0 9 6 , 6 8 8 2 6 …

28. 4 , 9 , 8 , 9 , 1 , 7 , 9 , 4 , 9 , 8 , 0 , 5 , 3 , 8 , 6 , 1 ,
9 , 6 , 3 , 8 , 9 , 1 , 3 , 7 , 6 , 0 , 8 , 0 , 5 , 8 , 8 , 8 ,
2 , 9 , 5 , 0 , 3 , 5 , 5 , 2 , 8 , 4 , 6 , 3 , 0 , 2 , 1 , 9 ,
6 , 1 , 9 , 1 , 6 , 1 , 4 , 1 , 4 , 5 , 3 , 6 , 3 , 3 , 6 , 8 ,
8 , 6 , 4 , 4 , 1 , 9 , 4 , 2 , 3 , 8 , 4 , 6 , 8 , 6 , 5 , 6 ,
4 , 9 , 7 , 7 , 3 , 9 , 0 , 0 , 9 , 4 , 2 , 9 , 0 , 8 , 0 , 4 ,
0 , 0 , 4 , 6 , 0

29. 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 0 , 1 , 1 , 1 , 1 ,
1 , 1 , 1 , 1 , 1 , 2 , 2 , 2 , 2 , 2 , 3 , 3 , 3 , 3 , 3 , 3 ,
3 , 3 , 3 , 3 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 , 4 ,
4 , 5 , 5 , 5 , 5 , 5 , 5 , 5 , 6 , 6 , 6 , 6 , 6 , 6 , 6 , 6 ,
6 , 6 , 6 , 6 , 6 , 7 , 7 , 7 , 7 , 8 , 8 , 8 , 8 , 8 , 8 , 8 ,
8 , 8 , 8 , 8 , 8 , 8 , 8 , 9 , 9 , 9 , 9 , 9 , 9 , 9 , 9 , 9 ,
9 , 9 , 9 , 9 , 9

30. for(IndexReader sub : reader.getSequentialSubReaders()) {
int p = 0;
for(int i = 0; i < MAX; i++) {
TermDocs td = sub.termDocs(new Term("fp", t));
int pos = td.read(docs, freqs);
while(pos != 0) {
for(int i = 0; i < pos; i++) {
if(p >= alld.length) {
alld = Arrays.copyOf(alld, alld.length * 2);
}
alld[p++] = docs[i];
}
pos = td.read(docs, freqs);
}
td.close();
}
if(p > 0) {
Arrays.sort(alld, 0, p);
int curr = alld[0];
int count = 0;
for(int i = 0; i < p; i++) {
int doc = alld[i];
if(doc == curr) {
count++;
} else {
nHits++;
curr += base;
heapCheck(h, hsize, curr, count);
curr = doc;
count = 1;
}
}
heapCheck(h, hsize, curr+base, count);
}
base += sub.maxDoc();
}
O ( n l o g n )

31. IntIntMap countMap = new IntIntMap(6000000, 0.9f);
for(IndexReader sub : reader.getSequentialSubReaders()) {
for(String t : termSet) {
TermDocs td = sub.termDocs(new Term("fp", t));
int pos = td.read(docs, freqs);
while(pos != 0) {
for(int i = 0; i < pos; i++) {
countMap.addOne(docs[i] + base);
if (docs[i] + base > max) {
max = docs[i] + base;
}
}
pos = td.read(docs, freqs);
}
td.close();
}
base += sub.maxDoc();
}
System.out.println("MAX: " + max);
int nHits = countMap.size();
int[] data = countMap.getData();
for (int i = 0; i < data.length - 1; i = i + 2) {
int key = data[i];
if (key != 0) {
heapCheck(h, hsize, key, data[i+1]);
}
}
O ( k n )

32. int[] countMap = new int[arraySize];
try {
for (IndexReader sub : reader.getSequentialSubReaders()) {
for (String t : termSet) {
TermDocs td = sub.termDocs(new Term("fp", t));
int pos = td.read(docs, freqs);
while (pos != 0) {
for (int i = 0; i < pos; i++) {
countMap[docs[i] + base]++;
}
pos = td.read(docs, freqs);
}
td.close();
}
base += sub.maxDoc();
}
} catch (ArrayIndexOutOfBoundsException e) {
increaseArraySize();
return eval(reader, queryTerms, rb, rows, start);
}
int nHits = 0;
for (int i = 0; i < countMap.length; i++) {
if (countMap[i] > 0){
nHits++;
heapCheck(h, hsize, i, countMap[i]);
}
}
O ( n )

33. P R E E M P T I B L E
Price reduction: $2000 or 66.6%
BEFORE AFTER
NUMBER OF SERVERS 6 2
COST / MONTH $3000 $1000

34. R I G H T S I Z I N G
R E C O M M E N D AT I O N S

35. E X T R A F O R S TA R T U S
• Home Office > co-working > Office
• Equity offers for developers
• Used equipment and furniture at auctions websites
• Government or universities programs (ex: startup Chile,
Startup Brazil, PIPE-FAPESP, USA NSF-ISBIR, Israel
OCS, etc.)
• Use cloud providers programs for startups to save costs
with infrastructure. Google, Rackspace, Microsoft,
Amazon, and others offer until US$100.000 for one year.
https://www.researchgate.net/publication/318429624

36. S TA R T U P C R E D I T S
https://cloud.google.com/developers/startups/

37. L U C K

38. B E C R E AT I V E

39. W H AT A B O U T Y O U ?
T W E E T T O @ D A N I C U K I W I T H # G C P S AV I N G S

40. D A7 F#m
Lembro quando a vida era difícil
F#m B7 Em A7
O servidor tinha vontade própria e dava pau
F#m B7 Em
O chefe reclamava o tempo inteiro
G Em A7 D
O senhor não vai gastar mais nenhum real
A7 D
Agora eu vou economizar
A7 A7add4 A7
Mudei pra nuvem
A7 G D
A minha alma hoje é meu celular
A7 A7add4
E todo o resto, todo resto
A7
eu dei pro Google
D
Eles que cuidem
D A7 D
Economizar,
D A7 G A7 D
com Google Cloud você vai economizar
D A7 D
Economizar,
D A7 G A7 D
com Google Cloud sei que vou economizar
D A7 D
Foi numa manhã de 6a. feira
D A7 D
Que me deu um ataque de fúria
B7 Em
Arremecei um extintor
A7 D
No armário do servidor
B7
E saí cortando
Em
todos os cabos de rede
A7 D
Explodi o roteador
B7 Em
Dei chute no monitor
A7 D
Joguei tudo na parede

41. “Nuvens… Hoje tenho consciência do céu, pois há dias em que
não o olho mas sinto, vivendo na cidade e não na natureza que a
inclui. Nuvens… São elas hoje a principal realidade, e preocupam-
me como se o velar do céu fosse um dos grandes perigos de meu
destino. Nuvens… Passam da barra para o Castelo, de Ocidente
para Oriente, num tumulto disperso e despido, branco às vezes, se
vão esfarrapadas na vanguarda de não sei o quê; meio-negro outras,
se, mais lentas, tardam em ser varridas pelo vento audível; negras de
um branco sujo, se, como se quisessem ficar, enegrecem mais da
vinda que da sombra o que as ruas abrem de falso espaço entre as
linhas fechadoras da casaria” Fernando Pessoa