Su'e le tele talafeagai mo se fuifui Kafka i Kubernetes

Manatua. fa'aliliu.: I totonu o lenei tusiga, o loʻo faʻasoa e Banzai Cloud se faʻataʻitaʻiga o le auala e mafai ai ona faʻaogaina ana meafaigaluega masani e faʻafaigofie ai le faʻaogaina o Kafka i totonu o Kubernetes. O faʻatonuga o loʻo i lalo o loʻo faʻaalia ai le auala e mafai ai ona e fuafuaina le tele sili ona lelei o au atinaʻe ma faʻapipiʻi Kafka lava ia e ausia ai le gaosiga manaʻomia.

Apache Kafka o se faʻasalalauga faʻasalalau faʻasalalau mo le fatuina o faiga faʻalagolago, faʻaleleia ma maualuga le faʻatinoina o taimi faʻafefe. E mafai ona fa'alauteleina ona agava'a e fa'aaoga ai Kubernetes. Mo lenei ua matou atiina ae Open Source Kafka operator ma se meafaigaluega e taʻua Supertubes. Latou te faʻatagaina oe e taʻavale Kafka i luga o Kubernetes ma faʻaoga ona foliga eseese, e pei o le faʻaleleia lelei o le faʻatulagaina o le au fai pisinisi, metric-based scaling ma le toe faʻaleleia, faʻalauiloaina o fata, "malulu" (matagofie) fa'asalalau fa'afouga, ma isi.

Taumafai Supertubes i lau fuifui:

curl https://getsupertubes.sh | sh и supertubes install -a --no-democluster --kubeconfig <path-to-eks-cluster-kubeconfig-file>

Pe fa'afeso'ota'i fa'amaumauga. E mafai foi ona e faitau e uiga i nisi o gafatia o Kafka, o le galuega lea e otometi e faʻaaoga ai le Supertubes ma le Kafka operator. Ua uma ona matou tusia e uiga ia i latou i luga o le blog:

• Oi leai! Ae o le isi Kafka operator mo Kubernetes;
• Mataʻituina ma faʻagaoioia Kafka e faʻavae i luga ole Prometheus metrics;
• Kafka fata fa'alauiloa ile Kubernetes;
• Ta'avale Apache Kafka i luga o Istio - fa'ailoga;
• E fa'amaonia e le tagata fa'aoga ma fa'aogaina fa'aputuga fa'atonutonu fa'atasi ma le Kafka operator;
• Kafka faʻaleleia le faʻaleleia ma le faʻamalosia o le faʻatulagaina i Kubernetes;
• Avefe'au protocol faamama mo Kafka, mashed.

A e filifili e faʻapipiʻi se faʻapipiʻi Kafka i luga o Kubernetes, e foliga mai o le a feagai oe ma le luʻitau o le fuafuaina o le tele sili ona lelei o atinaʻe faʻavae ma le manaʻoga e faʻalelei lau faʻatulagaga Kafka e fetaui ma manaʻoga. O le maualuga o le faʻatinoga o tagata fai pisinisi taʻitasi e faʻamoemoeina e ala i le faʻatinoina o vaega faʻavae autu, e pei o le manatua, gaioiga, saoasaoa o le disk, bandwidth network, ma isi.

O le mea e sili ona lelei, o le faʻatulagaina o tagata fai pisinisi e tatau ona faʻaogaina uma elemene atinaʻe i le maualuga o latou gafatia. Ae ui i lea, i le olaga moni o lenei seti e fai si lavelave. E foliga mai o le a faʻapipiʻi e tagata faʻaoga tagata faʻatau oloa e faʻateleina le faʻaogaina o se tasi pe lua vaega (tisiki, manatua, poʻo le gaosiga). I le tulaga lautele, e fa'aalia e le tagata fai pisinisi le maualuga o le fa'atinoga pe a fa'ataga e lona fa'atulagaina le vaega sili ona fa'agesegese e fa'aoga i lona tulaga atoatoa. O le auala lea e mafai ai ona tatou maua se manatu faigata o le avega e mafai e se tasi tagata fai pisinisi ona taulimaina.

Fa'ata'ita'iga, e mafai fo'i ona tatou fa'atatauina le aofa'i o tagata fai pisinisi e mana'omia e fa'afoeina se uta tu'uina atu. Ae ui i lea, i le faʻataʻitaʻiga o loʻo i ai le tele o filifiliga faʻatulagaina i tulaga eseese e matua faigata lava (pe a le mafai) ona iloilo le faʻatinoina o se faʻatulagaga faapitoa. I se isi faaupuga, e matua faigata lava ona fuafua se faʻatulagaga e faʻavae i luga o nisi faʻatinoga.

Mo tagata faʻaoga Supertubes, matou te masani ona faʻaogaina le auala lenei: matou te amata i nisi faʻatulagaga (infrastructure + settings), ona fuaina lea o lona faʻatinoga, fetuʻutuʻunaʻi faʻasologa o fefaʻatauaiga ma toe fai le faagasologa. E tupu lenei mea se'ia o'o ina fa'aoga atoatoa le vaega aupito telegese ole atina'e.

I lenei auala, matou te maua ai se manatu manino pe fia le aofaʻi o tagata faʻatau oloa e manaʻomia e le fuifui ona taulimaina se uta patino (o le numera o le au fai pisinisi e faʻalagolago foi i isi mea, e pei o le laʻititi o numera o faʻasalalauga e faʻamautinoa ai le maufetuunaʻi, numera o le vaeluaga. taitai, ma isi). E le gata i lea, matou te maua le malamalama po'o fea vaega tetele e mana'omia ai le fa'alava i luga.

O lenei tusiga o le a talanoa e uiga i laʻasaga tatou te faia e maua ai le tele o vaega sili ona lemu i uluai faʻasalalauga ma fuaina le gaosiga o se fuifui Kafka. O se faʻatulagaga sili ona maufetu e manaʻomia ai le itiiti ifo ma le tolu tagata faʻatau pisinisi (min.insync.replicas=3), tufatufaina i sone avanoa eseese e tolu. Ina ia fetuutuunai, fuaina ma mataʻituina le atinaʻe o Kubernetes, matou te faʻaogaina la matou lava faʻatautaia o pusa mo ao faʻapipiʻi - Pipili. E lagolagoina i luga ole fale (uamea, VMware) ma ituaiga e lima o ao (Alibaba, AWS, Azure, Google, Oracle), faʻapea foʻi ma soʻo se tuufaatasiga o ia mea.

Manatu i luga ole fausaga ma le faʻatulagaina o Kafka

Mo faʻataʻitaʻiga o loʻo i lalo, na matou filifilia le AWS e avea ma ao e tuʻuina atu ma EKS e avea ma tufatufaga Kubernetes. E mafai ona faʻatinoina se faʻatulagaga tutusa e faʻaaoga ai P.K.E. - Kubernetes tufatufaina mai Banzai Cloud, faʻamaonia e CNCF.

Disk

Amazon e ofoina atu eseese ituaiga voluma EBS. I le totonugalemu gp2 и io1 o loʻo i ai taʻavale SSD, peitaʻi, e faʻamautinoa le maualuga o le gaosiga gp2 fa'aaogaina aitalafu fa'aputuina (I/O credits), o lea na matou fiafia ai i le ituaiga io1, lea e ofoina atu fa'atasi le maualuga o le gaosiga.

Ituaiga fa'ata'ita'iga

O le faʻatinoga a Kafka e faʻalagolago tele i le faʻaogaina o itulau o le itulau, o lea matou te manaʻomia ai faʻataʻitaʻiga e lava le manatua mo le au fai pisinisi (JVM) ma itulau cache. Fa'ata'ita'iga c5.2xtele - o se amataga lelei, talu ai e 16 GB o le manatua ma sili ona lelei e galulue ma EBS. O lona fa'aletonu o le na'o le mafai ona tu'uina atu le maualuga o fa'atinoga mo le sili atu i le 30 minute i 24 itula uma. Afai e mana'omia e lau galuega mamafa le fa'atinoina o galuega i se vaitaimi umi atu, atonu e te mana'o e mafaufau i isi ituaiga fa'ata'ita'iga. O le mea tonu lena na matou faia, o le tu i c5.4xtele. E maua ai le maualuga o le gaosiga i totonu 593,75 Mbps. Tulaga maualuga o le EBS voluma io1 maualuga atu nai lo le faataitaiga c5.4xtele, o le mea lea o le elemene sili ona tuai o le atinaʻe e foliga mai o le I/O faʻaogaina o lenei ituaiga faʻataʻitaʻiga (lea e tatau foi ona faʻamaonia e a tatou suʻega uta).

Fesootaiga

Ole fa'aogaina ole feso'ota'iga e tatau ona lava tele pe a fa'atusatusa i le fa'atinoga o le fa'ata'ita'iga VM ma le tisiki, a leai o le feso'ota'iga e avea ma fagu. I la matou tulaga, o le fesoʻotaʻiga fesoʻotaʻiga c5.4xtele lagolagoina le saoasaoa e oʻo atu i le 10 Gb / s, lea e sili atu le maualuga nai lo le faʻaogaina o le I / O o se faʻataʻitaʻiga VM.

Broker Deployment

E tatau ona fa'apipi'i tagata fai pisinisi (fa'atulagaina i Kubernetes) i nodes fa'apitoa e aloese ai mai le tauva ma isi faiga mo le PPU, manatua, feso'ota'iga, ma puna'oa tisiki.

Java version

O le filifiliga talafeagai o le Java 11 aua e fetaui ma Docker i le uiga o le JVM e saʻo le fuafuaina o gaioiga ma manatua e avanoa i le atigipusa o loʻo tamoe ai le tagata fai pisinisi. O le iloaina o tapula'a o le PPU e taua tele, o le JVM i totonu ma manino fa'atulaga le numera o filo GC ma filo JIT. Na matou faʻaaogaina le ata Kafka banzaicloud/kafka:2.13-2.4.0, lea e aofia ai le Kafka version 2.4.0 (Scala 2.13) ile Java 11.

Afai e te manaʻo e aʻoaʻo atili e uiga i Java / JVM i Kubernetes, siaki matou pou nei:

• Aisea ua OOMKilled la'u talosaga Java;
• Fa'afefea ona sa'o le lapopoa o pusa mo Java 10 talosaga.

Seti manatua Broker

E lua itu taua i le faʻatulagaina o manatuaga a le au fai pisinisi: faʻatulagaina mo le JVM ma mo le Kubernetes pod. E tatau ona sili atu le tapula'a manatua mo se pod nai lo le maualuga o le fa'aputuga ina ia maua ai e le JVM le avanoa mo le Java metaspace, o lo'o nofo i lona lava manatua, ma mo le fa'aogaina o le upega tafa'ilagi, lea e fa'aaogaina e Kafka. I a matou suʻega na matou faʻalauiloaina Kafka brokers ma faʻamau -Xmx4G -Xms2G, ma le tapulaa manatua mo le pod o 10 Gi. Fa'amolemole maitau e mafai ona maua otometi le fa'aogaina o fa'amaumauga mo le JVM -XX:MaxRAMPercentage и -X:MinRAMPercentage, fa'avae i luga o le fa'agata manatua mo le pod.

Fa'atonuga fai pisinisi

I se tulaga lautele, e mafai ona e faʻaleleia le faʻatinoga e ala i le faʻateleina o le tutusa e ala i le faʻateleina o numera o filo faʻaaogaina e Kafka. O le tele o gaioiga e avanoa mo Kafka, o le sili atu lea. I la matou suʻega, na matou amata i se tapulaʻa o le 6 processors ma faasolosolo malie (e ala i le faʻasologa) siitia a latou numera i le 15. E le gata i lea, matou te setiina. num.network.threads=12 i totonu o le aufaipisinisi e faʻateleina le numera o filo e maua faʻamatalaga mai le fesoʻotaʻiga ma auina atu. O le taimi lava na iloa ai e le mafai e le au fai pisinisi mulimuli ona vave maua ni kopi, na latou siitia num.replica.fetchers i le 4 e faʻateleina ai le saoasaoa lea na toe faia ai e le au fai fefaʻatauaʻiga savali mai taʻitaʻi.

Meafaigaluega Fausia uta

E tatau ona e faʻamautinoa e le uma le malosi o le gaosiga o uta ua filifilia aʻo leʻi oʻo le fuifui Kafka (lea o loʻo faʻataʻitaʻiina) i lona uta maualuga. I se isi faaupuga, e manaʻomia le faia o se suʻesuʻega muamua o le gafatia o le gaosiga o meafaigaluega, ma filifili foi ituaiga faʻataʻitaʻiga mo ia ma le lava numera o processors ma manatua. I lenei tulaga, o la matou meafaigaluega o le a maua ai le tele o uta nai lo le kafka cluster e mafai ona taulimaina. Ina ua uma le tele o suʻesuʻega, sa matou faʻamautu i kopi e tolu c5.4xtele, o ia mea taitasi sa i ai se afi afi e ola.

Fa'atatauga

O le fuaina o faʻatinoga o se faʻagasologa faʻasolosolo e aofia ai vaega nei:

• fa'atūina mea tetele (EKS cluster, Kafka cluster, meafaigaluega fa'atupuina uta, fa'apea fo'i ma Prometheus ma Grafana);
• fa'atupuina o se uta mo se vaitaimi fa'apitoa e fa'amama ai fa'alavelave fa'afuase'i i fa'ailoga fa'atinoga ua aoina;
• fetu'una'iga o atina'e ma fa'atonuga a le faioloa e fa'atatau i fa'ailoga o fa'atinoga ua matauina;
• toe fai le fa'agasologa seia o'o ina maua le tulaga mana'omia o le fa'aogaina o le fa'aputuga o Kafka. I le taimi lava e tasi, e tatau ona toe faʻaleleia ma faʻaalia ni suiga laiti i le gaosiga.

O le vaega e sosoo ai o loʻo faʻamatalaina ai laʻasaga na faia i le taimi o le faʻataʻitaʻiga o faʻataʻitaʻiga.

Meafaigaluega

O meafaigaluega nei na faʻaaogaina e faʻapipiʻi vave ai se faʻatulagaga faʻavae, faʻatupu avega, ma fuaina le faʻatinoga:

• Banzai Cloud Pipeline mo le fa'atulagaina o se fuifui EKS mai le Amazon c Prometheus (ia aoina le Kafka ma mea tetele metrics) ma tusifana (ia vaai faalemafaufau i nei fua faatatau). Sa matou faaaogaina tu'ufa'atasi в Pipili 'au'aunaga e tu'uina atu ai le mata'ituina fa'atasi, fa'atotonugalemu o fa'aputuga o ogalaau, su'esu'ega fa'aletonu, toe fa'aleleia o fa'alavelave, puipuiga o atina'e ma le tele o isi mea.
• Sangrenel - o se meafaigaluega mo le suʻeina o uta o se fuifui Kafka.
• Grafana dashboards mo le vaʻaia o le Kafka metrics ma atinaʻe: Kubernetes Kafka, Node Exporter.
• Supertubes CLI mo le auala sili ona faigofie e faʻapipiʻi ai se fuifui Kafka i Kubernetes. Zookeeper, Kafka operator, Envoy ma le tele o isi vaega o loʻo faʻapipiʻiina ma faʻapipiʻi lelei e faʻatautaia se faʻaputuga Kafka saunia i luga o Kubernetes.
  • Mo faapipiiina supertubes CLI faaaoga faatonuga ua tuuina atu iinei.

EKS fuifui

Saunia se fuifui EKS fa'atasi ai ma pona tagata faigaluega c5.4xtele i sone avanoa eseese mo pods ma Kafka brokers, faʻapea foʻi ma nodes tuuto mo le gaosiga o uta ma atinaʻe mataʻituina.

banzai cluster create -f https://raw.githubusercontent.com/banzaicloud/kafka-operator/master/docs/benchmarks/infrastructure/cluster_eks_202001.json

A maeʻa loa le faʻapipiʻi EKS ma faʻagaoioia, faʻatagaina lona tuʻufaʻatasia auaunaga mata'ituina - o le a ia faʻapipiʻiina Prometheus ma Grafana i totonu o se fuifui.

Vaega o le polokalama Kafka

Faʻapipiʻi vaega o le Kafka (Zookeeper, kafka-operator) i le EKS e faʻaaoga ai supertubes CLI:

supertubes install -a --no-democluster --kubeconfig <path-to-eks-cluster-kubeconfig-file>

fuifui Kafka

E ala i le le mafai, EKS faʻaaogaina voluma EBS o ituaiga gp2, o lea e te manaʻomia ai le fatuina o se vasega faʻapipiʻi ese e faʻavae i luga o voluma io1 mo Kafka fuifui:

kubectl create -f - <<EOF
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: fast-ssd
provisioner: kubernetes.io/aws-ebs
parameters:
  type: io1
  iopsPerGB: "50"
  fsType: ext4
volumeBindingMode: WaitForFirstConsumer
EOF

Seti le parakalafa mo tagata fai pisinisi min.insync.replicas=3 ma fa'apipi'i pusa fai pisinisi i nodes i sone avanoa eseese e tolu:

supertubes cluster create -n kafka --kubeconfig <path-to-eks-cluster-kubeconfig-file> -f https://raw.githubusercontent.com/banzaicloud/kafka-operator/master/docs/benchmarks/infrastructure/kafka_202001_3brokers.yaml --wait --timeout 600

Mataupu

Sa matou fa'atautaia fa'atutusa afi afi e tolu. E tusi uma i latou i la latou lava autu, o lona uiga, tatou te manaʻomia ni autu se tolu i le aofaʻi:

supertubes cluster topic create -n kafka --kubeconfig <path-to-eks-cluster-kubeconfig-file> -f -<<EOF
apiVersion: kafka.banzaicloud.io/v1alpha1
kind: KafkaTopic
metadata:
  name: perftest1
spec:
  name: perftest1
  partitions: 12
  replicationFactor: 3
  retention.ms: '28800000'
  cleanup.policy: delete
EOF

supertubes cluster topic create -n kafka --kubeconfig <path-to-eks-cluster-kubeconfig-file> -f -<<EOF
apiVersion: kafka.banzaicloud.io/v1alpha1
kind: KafkaTopic
metadata:
    name: perftest2
spec:
  name: perftest2
  partitions: 12
  replicationFactor: 3
  retention.ms: '28800000'
  cleanup.policy: delete
EOF

supertubes cluster topic create -n kafka --kubeconfig <path-to-eks-cluster-kubeconfig-file> -f -<<EOF
apiVersion: kafka.banzaicloud.io/v1alpha1
kind: KafkaTopic
metadata:
  name: perftest3
spec:
  name: perftest3
  partitions: 12
  replicationFactor: 3
  retention.ms: '28800000'
  cleanup.policy: delete
EOF

Mo autu ta'itasi, o le fa'atusa e 3—le tau fa'atauva'a fa'atauva'a mo faiga fa'akomepiuta avanoa.

Meafaigaluega Fausia uta

Na matou faʻalauiloaina ni kopi se tolu o le gaosiga o uta (tusi taʻitasi i se autu eseese). Mo uta generator pods, e te manaʻomia le setiina o le node affinity ina ia faʻatulagaina naʻo luga o nodes ua tuʻuina atu mo i latou:

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  labels:
    app: loadtest
  name: perf-load1
  namespace: kafka
spec:
  progressDeadlineSeconds: 600
  replicas: 1
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      app: loadtest
  strategy:
    rollingUpdate:
      maxSurge: 25%
      maxUnavailable: 25%
    type: RollingUpdate
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: loadtest
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: nodepool.banzaicloud.io/name
                operator: In
                values:
                - loadgen
      containers:
      - args:
        - -brokers=kafka-0:29092,kafka-1:29092,kafka-2:29092,kafka-3:29092
        - -topic=perftest1
        - -required-acks=all
        - -message-size=512
        - -workers=20
        image: banzaicloud/perfload:0.1.0-blog
        imagePullPolicy: Always
        name: sangrenel
        resources:
          limits:
            cpu: 2
            memory: 1Gi
          requests:
            cpu: 2
            memory: 1Gi
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      terminationGracePeriodSeconds: 30

O nai manatu e matauina:

• O le gaosiga o uta e maua ai feʻau o le 512 paita le umi ma faʻasalalau i Kafka i vaega o feʻau 500.
• Fa'aaogaina o se finauga -required-acks=all O le faʻasalalauga e manatu e manuia pe a maua uma faʻasologa faʻasologa o le feʻau ma faʻamaonia e Kafka brokers. O lona uiga i totonu o le tagavai sa matou fuaina ai e le gata i le saoasaoa o taitai e mauaina savali, ae faapena foi i latou mulimuli i le toe faia o savali. O le fa'amoemoega o lenei su'ega e le o le su'esu'eina o le saoasaoa faitau tusi a tagata (tagata fa'atau) na maua talu ai nei savali o loo tumau pea i le OS itulau cache, ma lona faatusatusaga i le saoasaoa faitau o savali o loo teuina i luga o le tisiki.
• E 20 tagata faigaluega e fa'atutusa (-workers=20). O tagata faigaluega taʻitasi e aofia ai le 5 tagata gaosiga e faʻasoa le fesoʻotaʻiga a le tagata faigaluega i le fuifui Kafka. O le iʻuga, o afi taʻitasi e 100 tagata gaosi oloa, ma latou te lafoina uma feʻau i le vaega o Kafka.

Mata'ituina le soifua maloloina o le fuifui

I le taimi o su'ega uta o le fuifui o Kafka, sa matou mata'ituina fo'i lona soifua maloloina ina ia mautinoa e leai ni toe fa'afouina o le pod, leai ni fa'ata'ita'iga e le'i fa'aogaina, ma le maualuga o le gaosiga ma sina suiga laiti:

• E tusia e le generator uta ni fa'amaumauga masani e uiga i le aofa'i o fe'au fa'asalalau ma le fua o mea sese. E tatau ona tumau pea le fua o mea sese 0,00%.
• Faʻatonutonu vaʻa, faʻapipiʻiina e le kafka-operator, e tuʻuina atu se dashboard lea e mafai ai foi ona tatou mataʻituina le tulaga o le fuifui. Ina ia matamata i lenei laulau fai:

  supertubes cluster cruisecontrol show -n kafka --kubeconfig <path-to-eks-cluster-kubeconfig-file>

• ISR tulaga (numera o fa'atusa o le “in-sync”) fa'aitiitia ma fa'alautele e tutusa ma le 0.

I'uga o fua

3 tagata fai pisinisi, tele feʻau - 512 bytes

Faatasi ai ma vaeluaga tutusa tufatufaina i luga o le tolu tagata fai pisinisi, na mafai ona matou ausia le faatinoga ~500 Mb/s (pe tusa ma le 990 afe fe'au i le sekone):

O le faʻaaogaina o le mafaufau ole masini komepiuta JVM e le sili atu i le 2 GB:

Na o'o atu le fa'aogaina o le pusi I/O i luga ole laiga uma e tolu na fa'atautaia e le au fai pisinisi:

Mai faʻamatalaga i luga o le faʻaogaina o mafaufauga e ala i nodes, e mulimuli mai o le faʻaogaina o le polokalama ma le faʻaogaina na ave ~ 10-15 GB:

3 tagata fai pisinisi, tele feʻau - 100 bytes

A'o fa'aitiitia le tele o fe'au, e pa'ū le aofa'i e tusa ma le 15-20%: o le taimi e fa'aalu i le fa'agaioina o fe'au ta'itasi e a'afia ai. E le gata i lea, ua toetoe lava faaluaina le uta o le processor.

Talu ai o loʻo i ai pea ni pusa e leʻi faʻaaogaina, e mafai ona faʻaleleia le faʻatinoga e ala i le suia o le faʻatulagaga Kafka. E le o se galuega faigofie, ina ia faʻateleina le gaosiga e sili atu le galue ma feʻau tetele.

4 tagata fai pisinisi, tele feʻau - 512 bytes

E faigofie ona e faʻateleina le faʻatinoga o se faʻaputuga Kafka e ala i le faʻaopoopoina o tagata faʻatau fou ma tausia le paleni o vaeluaga (e faʻamautinoa ai o loʻo faʻasoa tutusa le uta i le va o tagata fai pisinisi). I la matou tulaga, ina ua uma ona faʻaopoopoina se tagata faʻatau, na faʻateleina le aofaʻi o fuifui i ~580 Mb/s (~1,1 miliona fe'au ile sekone). O le tuputupu aʻe na foliga mai e itiiti ifo nai lo le mea na faʻamoemoeina: e masani lava ona faʻamatalaina e le le paleni o vaeluaga (e le o tagata fai pisinisi uma e galulue i le pito i luga o latou gafatia).

O le taumafaina manatua ole masini JVM na tumau i lalo ole 2 GB:

O le galuega a tagata fai pisinisi ma taʻavale na aʻafia i le le paleni o vaeluaga:

sailiiliga

O le faʻataʻitaʻiga o loʻo tuʻuina atu i luga e mafai ona faʻalauteleina e aofia ai faʻaaliga sili atu ona faigata e aofia ai le faitau selau o tagata faʻatau, toe faʻavasegaina, faʻafouina faʻafouga, toe amata pod, ma isi. O nei mea uma e mafai ai ona tatou iloiloina tapulaʻa o le gafatia o le Kafka cluster i tulaga eseese, faʻailoa fagu i lana gaioiga ma saili auala e faʻafefe ai.

Na matou mamanuina Supertubes e vave ma faigofie ona faʻapipiʻi se fuifui, faʻapipiʻi, faʻaopoopo / aveese tagata fai pisinisi ma autu, tali atu i faʻasalalauga, ma faʻamautinoa e galue lelei Kafka i Kubernetes. O la matou sini o le fesoasoani lea ia te oe e taulai atu i le galuega autu ("fausia" ma "faʻaaogaina" savali Kafka), ma tuʻu uma galuega faigata i Supertubes ma le Kafka operator.

Afai e te fiafia i Banzai Cloud tekinolosi ma Open Source poloketi, faʻasoa i le kamupani ile GitHub, LinkedIn poʻo Twitter.

PS mai faaliliu

Faitau foi i la matou blog:

• «Tasi tala ma le Redis operator i K8s ma se iloiloga laʻititi o mea aoga mo le suʻeina o faʻamatalaga mai lenei faʻamaumauga";
• «Femalagaa'i sologa lelei a RabbitMQ i Kubernetes";
• «zetcd mai CoreOS: Suia le ZooKeeper ile...etcd storage".

puna: www.habr.com